Power and Marxist Theory: A Realist View

Power and Marxist Theory A REALIST VIEW

Jeffrey C. Isaac

Cornell University Press ITHACA AND LONDON

Copyright © 1987 by Cornell University All rights reserved. Except for brief quotations in a review, this book, or parts thereof, must not be reproduced in any form without permission in writing from the publisher. For information, address Cornell University Press, 124 Roberts Place, Ithaca, New York 14850. First published 1987 by Cornell University Press. International Standard Book Number O~8014-1934-4 Library of Congress Catalog Card Number 87-6690 Printed in the United States of America Librarians: Library of Congress cataloging information appears on the last page of the book.

The paper in this book is acid-free and meets the guidelines for permanence and durability of the Committee on Production Guidelines for Book Longevity of the Council on Library Resources.

For Debra and Adam, with love

Contents

Acknowledgments Introduction: On the Reason for Yet Another Argument about the Concept of Power

IX

PART I EMPIRICISM, REALISM, AND POWER

15

1. Behavioralism and the Three Faces of Power

17

Behavioralism and Empiricism 17 The First Face of Power: Behavioral Compliance 25 The Second Face of Power: Variations on the Same Theme

The Third Face of Power

1

29

33

Beyond the Three Faces of Power

39

41

2. Realism and Social Scientific Theory Realism, PostwPositivism, and Scientific Theory 42 Realism, Human Agency, and Social Practice 51 Realism and Social Structure 56 Criticism and Social Scientific Knowledge 63

72

3. The Concept of Power Revisited Power as Capacity 73 Power and Agency 75 Social Power and Social Structure 79 Power and Domination 83 Reciprocity and the Negotiation of Power Power, Interests, and Ideology 95

87

109

PART II MARXISM AND POWER

116

4. Marxism, Class, and Power Class and the Structure of Capitalism in Recent Writings Marxism and Class Powers 122

116 VII

Vlll

Contents

Class Formation: Agency, Struggle, and Collective Organization Objections Overturned: or What a Theory of Power Cannot Do

150

5. Marxism and the State The Poulantzas-Miliband Debate

152

Realism and the State 156 The Relative Autonomy of the Capitalist State The State as a Determinant of Class Relations Capitalism and the Determination of the State Class Struggle, Political Conflict, and the State

163 169 178 184

193

Militarism, Bureaucracy, and the State 198 Post-Marxism and the New Social Movements Marxism and the Limits of Power 219

Index

Acknowledgments 192

6. Power and the Limits of Marxism Pluralism and Ideology

132 140

208

233

During the time I have been working on this project I have incurred intellectual debts to many colleagues and friends. Three individuals must be singled out. lowe a special debt to Robert Dahl. As my teacher and as the supervisor of my doctoral dissertation at Yale, Professor Dahl has been a source of great inspiration as well as constant encouragement. Though my book is in large part a straightforward critique of his own work, Professor Dahl has always been its staunchest supporter. He has approached

my work with a spirit of open-mindedness and tolerance which is the hallmark of genuine scientific inquiry. At times I left his office in wonderment, having just spent hours discussing with him this writer named "Dahl" in the third person. I have learned many things from him, about the character of democratic states and about the character of democratic individuals. My undergraduate philosophy teacher at Queens College and my good friend and colleague ever since, Peter Manicas has been the primary intellectual influence on this book. Peter first introduced me to realist philosophy of science when he mailed me an essay by Roy Bhaskar back in 1979; and Peter's own prolific writing has been quite central in shaping my thinking on philosophical and political matters. He has read each chapter of this book assiduously, and he has provided me with numerous helpful suggestions. We have spent many hours arguing about realism and about Marxism and, though we have not always agreed, I could not have written the book without his encouragement and his questioning. I also gratefully acknowledge the influence of Raymond Franklin. IX

x

Acknowledgments

Ray first sparked my interest in political economy and Marxism. As an undergraduate at Queens College, I spent hours too numerous to count in his office, where I learned what real intellectual conversation and real intellectual kinship were about. Other people have helped me in the course of my work. James Farr, Terence Ball, Alan Gilbert, Paul Thomas, G. William Domhoff, and Allen Graubard all read the entire manuscript and offered valuable comments and criticisms. (Special thanks go to Jim, who read it all twice.) Roy Bhaskar, David Cameron, David Johnston, Michael Krasner, Irving Leonard Markovitz, David Mayhew, Douglas Rae, Gerald Rosenberg, Ian Shapiro, David Sprintzen, Alexander Wendt, Robert Paul Wolff, Erik Olin Wright, and Burt Zweibach read parts of the manuscript and made helpful comments. Three friends and colleagues at Fordham, Patricia Clough, Barry Goldberg, and Robert Orsi, have provided me with many stimulating conversations during the writing of this book, and the insights gleaned from these have undoubtedly made their way into the final draft. I also thank John Ackerman of Cornell University Press, editors Kay Scheuer and Lois Krieger, and Deborah Hause, who have helped turn my manuscript into a better book. Finally, I thank my family for their help and support throughout the writing of this book. My parents, Sylvia and Hyman Isaac, and my brother, Gary Isaac, have always encouraged me. From my father, a linotype operator, I first learned of the nobility of work, and of its degradation under capitalism; and it was over the dinner table that I first learned the value of political argument. My wife, Debra Kent, a writer herself, has provided me with more support, emotional, intellectual, and financial, than any scholar has any right to expect. To her, and to our new son Adam, this book is dedicated. JEFFREY

Farmingdale, New York

C. ISAAC

Power and Marxist Theory

INTRODUCTION

On the Reason for Yet Another Argument about the Concept of Power Social theorists seem to have an unfailing inclination toward methodological controversy. Such argument fills the pages of professional journals, and it makes (and breaks?) careers. But there is, we suspect, more to it than this. And so we continue the argument. Ever since Aristotle, it has seemed impossible to write about society. without also writing about writing about society. Sheldon Wolin has suggested that there is a political point behind such a concern with method, that argument about method is also both argument about theory and argument about the good society.' If this is true in general, it would seem to be supremely true of contemporary social science. We live in an age of science and technology. And even if the environmental horrors of the past two decades have caused us to reassess their value, there can be no doubt that understandings of science, and claims to scientificity, have played a crucial role in the development of contemporary social theory. This is particularly true of recent discussions of the concept of power, in which notions of causality and proper scientific explanation have, as I shall argue, structured debate within certain narrow limits. There is a justifiable inclination to circumvent all this discussion, to extricate oneself from the tangled conceptual webs that social scientists have spun, and simply to do actual research. lSee Sheldon S. Wolin, "Max Weber: Legitimacy, Method, and the Politics of Theory," Political Theory 9 (August 1981),401-24.

1

2

Introduction

The proof of the pudding, after all, is in the eating, isn't it? Unfortunately, things are not that easy, for the question of whether we are properly eating the pudding, and whether it is indeed pudding that we are eating, will not go away. The conceptual problems remain, and they determine our scholarly activity. What, after all, is "actual research"? And what is this thing called "power" that we are researching?

Pure meta-theoretical argument tends toward a hopeless abstractness, and despite the important contribution of the "linguistic turn" in contemporary philosophy, which has taught us that we cannot get outside of our language, 2 the utilitarian in us forces us to ask for the theoretical payoff-is there a point, or are we simply spinning conceptual webs? In spite of this danger, however, one suspects that the injunction simply to "do research" is based on . an extremely naive view of what constitutes scholarly activity.' Political science, like any science, is a social practice. It is an activity undertaken by human beings in certain definite relationships and governed by norms, rules, and understandings that make activity meaningful. This may seem like an incredibly obvious and trivial claim with which to begin a book. But it is only recently that this sort of claim has become an accepted piece of commonsense wisdom among social theorists. Social scientists, particularly in America and particularly since the 1930s, have until recently believed that their endeavors involved primarily the observation and logical interpretation of a set of un mediated "facts," given in experience and fundamentally unproblematical. It is not necessary to rehearse the development of this self-understanding, centrally informed by logical positivist philosophy, nor is it necessary to rehearse its demise. It is sufficient to note that theorists as different as Thomas Kuhn, Jiirgen Habermas, W.V.O. Quine, and Michel Foucault have taught us that as social scientists our encounters 2See Richard Rarty, Philosophy and the Mirror of Nature (Princeton: Princeton University Press, 1979). 3For the most sophisticated but, in my view, flawed polemic against methodological analysis, see John G. Gunnell, "In Search of the Political Object: Beyond Methodology and Transcendentalism," in John S. Nelson, ed., What Should Political Theory Be Now? (Albany: State University of New York Press, 1983),

pp.25-52.

Introduction

3

with "reality" are fundamentally constituted by the normative and conceptual frameworks within which we operate. 4 This understanding is absolutely critical from an epistemological point of view. It means that our theories about society are complex webs of interrelated concepts, and that there can be no Archimedean point from which to "observe" the world and adjudicate the "Truth" of scientific theories. The importance of this point, however, extends beyond the now familiar recognitions that facts are theory-laden and truth is rule-bound and irreducibly judgmental. For the nOrms and preconceptions governing social scientific practice not only (partly) determine our substantive conception of the world, they also determine our very notions of the proper methods of studying the world and, more generally, our understanding of its ultimate constitution, its ontology. In this book I am not primarily interested in analyzing the set of substantive concepts that constitutes the dominant theory about power in political science-pluralism. This task is, of course, a very important one, one to which many others who figure in this book have set themselves. William Connolly's Political Science and Ideology (1967) and The Bias of Pluralism (1969), for example, were early efforts to expose the complex descriptive and normative presuppositions of pluralist theorists and their critics. And Peter Bachrach and Morton Baratz, in their Power and Poverty (1970), emphasized these issues in their critique of pluralism. However, while much of this important work staked its claim on the terrain of epistemology, opening up space for the critique of pluralism by employing post-positivist notions of theory and its irreducibly interpretative dimension, it never really challenged the consensus on other methodological issues, particularly ontological ones. Of course some of these questions, regarding the nature of social being, intersubjectivity, and human agency, were not wholly ignored by political science. As Richard Ashcraft has noted, they were taken up by those who extricated themselves from the methods of behavioralism and asserted their independence as political philosophers and political theorists rather than as political scien4See Richard J. Bernstein, Beyond Objectivism and Relativism: Science, Her~ meneutics, and Praxis (Philadelphia: University of Pennsylvania Press, 1983),

4

Introduction

tists. 5 Here, these inquiries became part of the discipline of textual hermeneutics, focused on the various issues surrounding the reading of the classic texts of the Western tradition. But even here, ironically, a certain consensus about the nature of science, regarding the nature of causality, was hardly challenged. Thus, a rough division of lahar developed in political science. On the one hand were those political scientists involved in actual, empirical research who, regardless of their avowed epistemology, undertook to do "real science" by arriving at lawlike predictive generalizations about political behavior. On the other hand were those who, repelled by this scientism, virtually abandoned causal explanation in favor of historical, exegetical, and moral inquiry. In this book I will analyze those methodological, or metatheoretical, understandings that are shared by most political scientists, pluralist theorists and their critics. In doing so, I hope to suggest, implicitly, the folly of the division of labor between political science and political theory which has developed in political inquiry. My primary interest, however, is in showing how social scientific research is shaped not only by substantive theoretical differences-pluralism versus elitism versus ~.1arxism versus feminism-but by more general norms regarding the nature of scientific inquiry. The so-called "three faces of power" controversy is a perfect example of the way social science is normatively constituted. All the contestants in this controversy share a set of understandings, which I will label empiricism; and these understandings have determined both the acceptable ways of thinking about social power and the unacceptable ways. This specific issue is particularly germane because of the perdurance and the irresolution of the controversy. Both of these are marked by the publication of the second edition of Nelson Polsby's Community Power and Political Theory in 1980.' The reissue of Polsby's book is interesting because the volume constitutes both a SRichard Ashcraft, "One Step Backward, Two Steps Forward: Reflections upon Contemporary Political Theory," in Nelson, ed., What Should Political Theory Be Now? pp. 515-48. See also my "After Empiricism: The Realist Alternative," in Terence Ball, ed., Idioms of Inquiry: Critique and Renewal in Political Theory (Albany: State University of New York Press, 1987). 6Nelson Poisby, Community Power and Political Theory, 2d ed. (New Haven: Yale University Press, 1980).

Introduction

5

recognition of the theoretical sterility of the entire debate about power and a restatement of old themes. Polsby notes the bankruptcy of the controversy but attempts to fortify his own position within it. The relevance of debate concerning the concept of power, however, is hardly limited to questions of the organization of political science as an intellectual discipline. The concept of power is central to all social scientific inquiry. It figures in contemporary debates regarding class structure, gender relations, the nature of the state, nuclear arms, and the danger of what Edward Thompson calls "exterminism." To locate the sources of power in society is to locate the enablements and constraints that operate on all of us as ordinary individuals, and that operate as well on those C. Wright Mills referred to as "extraordinary" individuals, those particularly powerful persons whose activities so shape our own. To locate power is also to fix moral responsibility, both upon those who exercise power illegitimately and upon those social structures that make this power available. To locate power is thus an enterprise central to "normative" as well as "descriptive" inquiry, for to do so is to identify the causes of what John Dewey called our "public problems," as well as the barriers to the solution of these prohlems. As Mills wrote: "The interest of the social scientist in social structure is not due to any view that the future is structurally determined. We study the structural limits of human decision in order to find points of effective intervention ... to find within them the ways in which they are and can be controlled. For only in this way can we come to know the limits and the meaning of human freedom. ,,7 My own interest in the presuppositions underlying the conceptualization of power come out of my experiences, as a graduate student, in confronting the "power" of Pols by's arguments, which have had the manifest effect of delegitimizing any talk of "power structure" or "dominant class." As a student of critical bent, interested in Marxian theory, I read Polsby, and came up squarely against the question of what the notion of a "dominant class" could possibly mean if not that one group regularly and predictably prevails over another under definite conditions. This road block was produced mainly by a set of sincerely held beliefs-some spe7C. Wright Mills, The Sociological imagination (Oxford: Oxford University Press, 1959), p. 174.

6

Introduction

cific (and rather primitive) notions of linguistic meaning, and some more general understandings of the nature of scientific explanation-and not by a set of coercive sanctions. My search for an answer led me to the works of Bachrach and Baratz, and of Steven Lukes, commonly taken to be the "radical" critics of Dahl and Polsby; I read their work with great excitement, but quickly became dissatisfied. Lukes seemed right that the approach of Bachrach and Baratz was little different from that of Dahl and Polsby. But while he raised some stimulating questions, Lukes's discussion, particularly his treatment of the problem of "objective interests," generated as many problems as it attempted to resolve. This debate helped me neither to understand the concept of power better nor to ground my belief that Marxism was a genuine theory of power. At the same time that I was dealing with these issues, I developed a growing interest in the body of work in philosophy and social theory dealing with the problems of the "received" view of social science. My absorption in this literature and my understanding of these issues, centrally informed by Roy Bhaskar's A Realist Theory of Science (1978),' led me to an insight into the three faces of power debate-that it centered around an empiricist view of causation and scientific explanation. Further inquiry led me to locate the limitations of this debate precisely in its empiricism, and to seek to formulate an alternative conceptualization of power. For all the participants in the three faces of power debate, power is a behavioral concept that refers to the regular conjunction of the behaviors of two parties, such that "A has power over B" means that in some sense A's doing something gets B to do something. This view is doubly confused. First, it is limited to situations of "'power over," and fails to see that "power over," or what I will call domination, is parasitic upon a "power to." Second, it fails to distinguish between the possession and the exercise of power. These confusions rest upon a refusal, characteristic of empiricism, to talk of the "capacities" and "intrinsic natures" of things. As Jack Nagel has put it, such talk has "objectionable .. metaphysical implications. ,,9 But as we shall see, the weaknesses of this debate are 8Roy Bhaskar, A Realist Theory of Science, 2d ed. (Atlantic Highlands, N.).: Humanities Press, 1978). "Jack Nagel, The Descriptive Analysis of Power (New Haven: Yale University Press, 1975), p. 36.

Introduction

7

themselves the result of certain objectionable implications of empiricist metaphysics-namely, the belief that the social world is nothing more than a set of behavioral regularities. Realist philosophy of science provides a different understanding of the social world, according to which society is a complex of structured relationships. If for empiricism the task of social science is the discovery and prediction of empirical regularities of the form "whenever A ... then B, for realism science analyzes those enduring mechanisms that cause the occurrence of empirical events. On this view a much more adequate concept of power, as socially structured capacity, is possible. This version of realism is to be distinguished from two other views that are also conventionally labeled "realist." The first is the doctrine of Realpolitik of the Kissinger/Morgenthau variety, a praxis according to which hardheaded "practicality" in a world of incessant struggle is recommended." Realist philosophy of science bears no necessary relationship to this doctrine. While the realist ontology I will articulate below shares with the doctrine of Realpolitik an emphasis on the constraints upon agents' choices, and while I will argue that power is a concept that marks this question of constraint, my version of philosophical realism in no way entails that aggrandizement and domination are transhistorical human imperatives. In fact, my version of realism insists upon the historical specificity of forms of social power and, in highlighting the possibility of the historical transformation of these forms, has critical implications for social practice which distinguish it from the generally conservative orientation of Realpolitik doctrines. The second version of realism is a view within the philosophy of science, which holds that scientific theories converge upon a rough correspondence with reality, and are thus progressively more realistic. This view is often associated with the view of realism developed in this book, and there are some genealogical affinities, particularly in the common criticism of positivism and idealism. This latter view of realism, however, in the view of many philosophers, occludes the significance of the social dimension of science and severely underestimates the complexity of scientific language and judgment and the discontinuity that characterizes the actual history of science. II

lOSee R. N. Berki, On Political Realism (London: Dent, 1981).

8

Introduction

Introduction

9

The view developed here, then, recognizes the necessary tension between the real objects of scientific explanation and theoretical explanation itself, and it posits no necessary correspondence between our understanding of the world and the way the world really operates. Thus realism here is not an epistemological claim about the progressive truth value of theories, but an ontological claim about the nature of causality and the form causal theories should properly take. This alternative, realist understanding of power is what is necessaty to ground Marxian theory, whose own analyses of power are not assimilable to the empiricist view. Marxian theory represents a different kind of theorizing about power than that envisioned by empiricism, a kind of theorizing in which the realist language of causal necessity and social structure plays a crucial . role. And in developing a realist view of power, I discovered that such a perspective illuminates some of the central concepts of Marxian theory: class domination, class capacities, class struggle, and the relative autonomy of the state. Illuminating Marxian theory, however, involves more than simply joining it to the realist view of power I formulate. It involves an extensive exposition and elaboration of its characteristic claims. This is necessary in order to combat the interpretation of Marxism proffered by mainstream political scientists and well articulated by Dahl, who writes that "according to Marxist theory, in capitalist societies, the capitalist class unilaterally rules over the society. In this sense the class as a whole might be considered the prime mover. But the unilateral dominance of the capitalist class is more a theoretical postulate or hypothesis than a wen verified description of contemporary non-socialist nation-states with 'democratic' governments."" On this view the Marxian theory of capitalism involves a global claim to the effect that capitalists unilaterally dominate, or are the prime mover-that is, they always get what they want. This, it should be noted, is the same view articulated by Karl Popper in his many critiques of Marxism-namely, that Marxism is a historicism, that it makes predictions about events (e.g., capitalists always prevail) without specifying the empirical

conditions under which these predicted events can be expected to occur. Because Marxian theory makes such categorical instead of couditional claims, because it talks about the nature of capitalism rather than a set of empirical uniformities, it is unscientific. 12 This kind of rejection of Marxism involves two sorts of claim. First it involves a specific theory of science--empiricism-which maintains that a scientific theory is a set of empirical predictions capable of falsification by the occurrence of a counterinstance. Second, it involves a particular interpretation of Marxian theory, according to which the notion of a dominant class involves a global prediction about events to the effect that capitalists always and unilaterally get what they want. In arguing that Marxism represents a genuinely scientific theory of power, I will challenge both of these beliefs. Part I of this book is a strictly meta-theoretical treatment of the questions of science, social science, and power. This section is a critique of the three faces of power debate. By critique I do not mean rejection, but rather a critical analysis that dissects a given reality (in this case, the reality of empiricist theory) and explains its weaknesses. My analysis is a critique in that it dissects the empiricist concept of power, explains its weaknesses in virtue of its empiricism, and offers an alternative conceptualization. I develop a realist understanding of social power, according to which power is implicated in the enduring structural relations that characterize a society and is exercised by intentional human agents who participate in these relations. This view of power, I argue, not only makes better sense of the concept; it also integrates an understanding of social structure and human agency in a way empiricist formulations cannot. Part II is a discussion of Marxian theory, particularly in its contemporary form, in the light of this argument. Here I argue that Marxism as a general orientation represents a scientific theory of social power, and that recent Marxian analyses of the class structure and politics of capitalist society can be understood only in realist terms. This section is not a consideration of the many important gaps and lacunae in Marxian theory, nor is it a substantive

llRobert A. Dahl, Modern Political Analysis, 3d ed. (Englewood Cliffs N.J.: Prentice-Hall, 1976), p. 41.

12See, for instance, Karl Popper, Conjectures and Refutations: The Growth of Scientific Knowledge (New York: Harper Torchbooks, 1965).

j

10

Introduction

defense of Marxism as a correct theory. It isa defense of Marxism as a genuine theory that deserves to be examined and criticized on its own merits as a theory, and not simply discarded on strictly formal (and specious) grounds because it fails to conform to the canons of empiricist philosophy. There is within Marxism a dangerous strand of essentialism, particularly as regards the Marxian theory of history, historical materialism. This essentialism is often expressed in reductionist accounts of social relations and political events, which would explain everything with reference to class relations and class interests. It is often expressed in what Louis Althusser has called "historicism," according to which history is treated as a unilinear process with an unfolding meaning and a necessary end-socialism. This essentialist Marxism would view gender conflicts, for instance, as being "really" the expressions of class conflict. And it frequently issues global predictions of the necessary instability of capitalism and the ruptural "moment" of socialist transformation. The version of Marxism I reconstruct here attempts to avoid these problems, by disavowing any necessary connection between the theory of capitalism and a theory of universal history~ and by concentrating exclusively on the former. Even this maneuver does not, however, insulate Marxian theory and practice from a number of serious criticisms. The point I would make about these weaknesses in Marxism, however, is that they are not a function of it not being properly scientific. They are a function of it being an incomplete, an inadequate, and possibly an untrue science. The point of this book is simply to argue that whatever defects exist within Marxism, they are not the defects charged by empiricism. It is not the fact that Marxism fails to conform to the deductivenomological view of scientific explanation that accounts for whatever problems confront it. It is not the fact that it employs a metaphysical language of structure, that it talks about the nature of capitalist society, and that it attempts to uncover causal mechanisms. If Marxism is an inadequate social theory, this is because it fails to specify the proper causal mechanisms. I thus do not claim that Marxism is the only realist scientific theory of power. In fact, I conclude by suggesting that pluralism represents an alternative theory, although its epistemological and ontological grounding is not the grounding its empiricist defenders

Introduction

11

believe it to be. Once again, however, to say pluralism is a genuinely scientific theoty is not to say that it is true. And so I also conclude by suggesting some of its deficiencies and by suggesting that, despite its limitations, Marxist theory is a necessary, if not a sufficient, condition of our understanding of contemporary society. This book is thus a normative argument, in the same way that Polsby's Community Power and Political Theory, Lukes's Power: A Radical View, and Nagel's The Descriptive Analysis of Power were normative-it makes an argument about how power should be studied, and more specifically, it attempts to justify the scientific status of Marxian theory. In the course of the discussion I quite self-consciously issue such injunctions; and, in doing so, I differ from my predecessors in this debate only in my open recognitio" of the normative implications of such an inquiry. Like my predecessors, I too proceed from an understanding of scientific practice to a discussion of the possibility of social scientific theories of power. Unlike them, I proceed from a different understanding of science, with different implications for social scientific practice. Also unlike them, in recognizing the lack of innocence of such a project, I attempt to deal with my theoretical opponents as generously as I can, directing my efforts primarily toward the legitimation of Marxian theory rather than the disqualification of its competitors. For as Roy Bhaskar has written: "There is no way in which philosophy can legislate in advance for the transposition of particular scientific procedures."" In other words, ultimately substantive considerations and judgments, and not abstract conceptual inquiry about power, are the only arbiters of social scientific truth. One more point is in order regarding the general intention underlying this book. The aspiration to scientific status has been an important regulative ideal of contemporary political theory. The rhetorical function of this book is to expose some of the problems with the dominant view of science and, in doing so, to argue bow social scientific theories of power should more properly be framed. I accept the ideal of scientific explanation, but only once it is shorn of its positivistic and honorific connotations. 13Roy Bhaskar, The Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences (Atlantic Highlands, N.J.: Humanities Press, 1979), p. 41.

12

Introduction

Science is not to be equated with the kind of exotic, highly formalized aud rarefied knowledge idealized by positivism. Some of the natural sciences do approximate this ideal in the kind of precision attained under laboratory conditions. But I hold, with Aristotle, that no science can demand more precision than its objects allow. And the objects of social science, the practices and relations of human beings, are implicated in the ongoing flow of interaction; they are constantly changing, constantly being reinterpreted by their participants. Social scientific theory can never get away from the f1esh-and-blood reality of social practice and historical process. This does not make our theories more impressionistic and less subject to critical scrutiny and empirical contestation. But it means that what we are studying is richly textured, and inherently meaningful and interpretative, in a way mere nature is not. This places limits on what social science can be. Its concepts necessarily draw upon lay concepts. And its theories are irremediably qualitative, referring to dimensions of social life that can not be closed off, measured, and predicted in the confines of a laboratory experiment. Neither is science a privileged form of knowledge, guaranteeing access to a special form of truth. Scientific knowledge, like all forms of knowledge, is fallible. And science, like all other human activities, is sustained by a social community engaged in a set of relations with the wider society. These realities point to the perennial question of the relation between theory and practice, or, more accurately, between the theoretical practice of science and the other, nonscientific practices that society is comprised of. This book is premised on the belief that social science can produce theoretical knowledge about society, knowledge that may not be apparent to ordinary social agents. This knowledge, however, is not privileged. It cannot negate the tacit skills and ordinary knowledge possessed by human beings. Nor can this knowledge make any necessary claim upon tbose it presumes to kuow about. Social science is a relatively autonomous practice. And the truth claims of its knowledge are properly valid only within the community of its practitioners. The truths (with a small "t") of social science do not license any practical applications. The truths of Marxism do not license a dictatorship of a privileged party elite. And tbe truths of policy study do not license the bureaucratic implementation of policy.

Introduction

13

Social science may have practical implications, but their practical application always requires the mediation of other forces and other considerations.

This is not an argument for value neutrality. Nor is it an argument for the kind of skeptical nihilism that seems to be so popular among many self-styled "postmodern" political theorists. It is the premise of this book that social science can uncover features of social life and properties of social relationships that are opaque to its participants. Social science thus has critical and emancipatory implications. This is what makes arguments about the concept of power and study about the structure of power so important. But we must recognize that the belief in the critical promise of social science is not a form of scientism, in which the truths of scientific inquiry run roughshod over all other considerations. In fact, this book, insofar as it is interested in power, is based on the diametrically opposed belief that social science must always come to terms with the social facts of power. For apologetic theorists this coming to terms is a very easy thing. For more critical theorists, the truths of their analysis must be expressed and articulated in such a way that they persuade those whom they propose to emancipate. The truths of social science never speak for themselves, or act by themselves, outside of the scientific community. They must be made practically relevant by nonscientific, political activity. In making our theories critically relevant in social and political affairs, we can do no better than to heed the words of C. Wright Mills: If we take the simple democratic view that what men are interested in is all that concerns us, then we are accepting the values that have been inculcated, often accidentally and often deliberately by vested interests ....

If we take the dogmatic view that what is to men's interests, whether they are interested in it or not, is all that need concern us morally, then we run the risk of violating democratic values. We may become manipulators or coereers, or both, rather than persuaders in a society in which men are trying to reason together and in which the value of reason is held in high esteem. 14

In avoiding the Scylla of political quiescence, we must also avoid the Charyhdis of a cynical Machiavellianism. The theoretical analJ4Mills, p. 194,

14

Introduction

ysis of power can play an emancipatory role. But if theory is to play this role it must be aware of its limits, and of the practical considerations that both make it relevant and temper its application.

PART I

EMPIRICISM, REALISM, AND POWER There have been too many books and articles written by social scientists on the concept of power. What follows in Part I is yet another intervention in this discourse. But it is one distinguished by its serious attention to recent developments in the philosophy of the social sciences, which have made it possible both to understand better the limits of mainstream academic discourse and to propose a more plausible alternative. The first, and more widely recognized, of these developments is the growing academic acceptance of the various philosophies Anthony Giddens has usefully labeled "interpretative sociologies." This general perspective, which has gained a particularly wide currency among political theorists, has made it possible to recognize the deficiencies of social scientific empiricism, and it has placed the questions of human agency and human understanding at the heart of social and political inquiry. The second, and less widely known, development is the convergence of many philosophers and social inquirers upon a realist understanding of scientific practice. If the first development has opened an unbridgeable gap between social study and what has been commonly understood by empiricists to be scientific inquiry, the second has more directly challenged this empiricist un-

15

16

Empiricism, Realism, and Power

derstanding of science itself. Realism has important implications for social and political analysis, and one of the underlying intentions of Part I is to make some of these clear. But the primary intention is to articulate a coherent concept of power, one that incorporates the insights of interpretative sociology but that refuses to abandon the ideal of scientific and critical knowledge.

CHAPTER 1

Behavioralism and the Three Faces of Power Intrafamily disagreements sometimes take on the color of major conflicts, family members in the heat of conflict forgetting the commonalities binding them together. Often it is only when confronted by an outsider that families under stress become aware of their relatedness and begin to recognize that what seemed like a major disagreement was really a family squabble. The three faces of power controversy in social theory is like a family squabble. A good deal of recrimination has been expressed by the debaters, who have understood their positions to represent major differences. When exposed to an outside perspective, however, the similarities between the contestants far outweigh any differences. Despite the differences between Dahl, Bachrach and Baratz, and Lukes, all three remain wedded to an empiricist understanding of science. They agree that power is a causal concept, but they understand causaliry as no more than a regular sequence of behaviors. This can be seen as the legacy of behavioralism in social science. The consequence of this legacy is that, while the "second" and "third" faces of power point to a structural conception, they are ultimately unable to sustain such a view.

Behavioralism and Empiricism The "behavioral revolution" in Anglo-American political science was a revolution in the methodology of practicing the academic discipline of political science. In response to what was believed by

17

18

Empiricism, Realism, and Power

some to be a wrongful preoccupation with historical, descriptive, and constitutional studies, behavioralism demanded a focus on the more "informal," nongovernmental political processes and a concern with greater "scientific rigor."l Behavioralism in political science has, not surprisingly, had important and long-term effects on the practice of political research. These effects, however, have been much less innocent than the revolutionary vanguard believed. Robert Dahl, in his famous "'monument to a successful protest," wrote:

"The behavioral approach is an attempt to improve our understanding of politics by seeking to explain the empirical aspects of political life by means of methods, theories, and criteria of proof that are acceptable according to the canons, conventions, and assumptions of modern empirical science.,,2 This quote gives something of the flavor of the intellectual moment-optimistic, naively self-assured about the nature of the "scientific outlook" that was to be emulated. As Dahl un-self-consciously makes clear, however, the triumph of this "protest movement" represented in fact much less the triumph of scientific methods than an emerging hegemony of an empiricist view of science. Dahl quotes from an early prescriptive tract, which he presents as simply and matter-of-factly scientific: "[We favor] a decision to explore the feasibility of developing a new approach to the study of political behavior. Based upon the study of individuals in political situations, this approach calls for the examination of the political relationships of men ... by disciplines which can throw light on the problems involved, with the object of formulating and testing hypotheses concerning uniformities of behavior."3

Dahl's essay makes clear that this emerging "scientific outlook," which hoped "to provide political science with empirical propo-

Behavioralism and the Three Faces of Power

19

sitions and theories of a systematic sort, tested by closer, more direct and rigorously controlled observations of political events," had strong political and bureaucratic determinants. 4 What is central for our present purposes, however, is that the view of scientific explanation as the documentation and prediction of empirical uniformities was a central tenet of behavioralism. David Easton has written that a theory is "any kind of generalization or proposition that asserts that two or more things, activities, or events, covary under specified conditions."'> A more recent and influential book on scientific method in political research asserts: "Science is concerned with the explanation (and prediction) of specific events by means of statements which are invariantly true from one set of circumstances to another.,,6

This conception of science is not peculiar to political science methodologists. Until very recently it has been the standard view in the philosophy of science. As Ernest Nagel wrote. in his authoritative The Structure of Science: "The sciences seek to discover and to formulate in general terms the conditions under which events of various sorts occur, the statements of such determining conditions being the explanations of the corresponding happenings.'" Science, on this view, consists of the documentation and prediction of empirical regularities. Scientific explanation consists of the sub4Ibid., p. 766. Dahl rather un~self-consciously celebrates the new state interest in financing research after World War II, and the role that academic institutions like the American Political Science Association and the Social Science Research Council played in supporting behavioral research. For a less sanguine view, one might consult C. Wright Mills, The Sociological Imagination (Oxford: Oxford University Press, 1959), particularly chap. 5, "The Bureaucratic Ethos"; see also Douglas Rae's "Political TIleory and the Division of Labor in Society: Asleep aboard the Titanic and Steaming into Halifax," Political Theory 9 (August 1981),

369-78. lFor an overview, see Bernard Crick, The American Science of Politics (Berkeley: University of California Press, 1959); Raymond Seidelman and Edward ]. Harpham, Disenchanted Realists: Political Science and the American Crisis, 18841984 (Albany: State University of New York Press, 1985); Heinz Euleau, The Behavioral Persuasion in Politics (New York: Random House, 1963); and David Truman, "Disillusion and Regeneration: The Quest for a Discipline," American Political Science Review 59 (December 1965), 865-73. 2Robert A. Dahl, "The Behavioral Approach in Political Science: Epitaph for a Monument to a Successful Protest," American Political Science Review 58 (De-

cember 1961), 767. 'Ibid., p. 764.

sDavid Easton, A Systems Analysis of Political Life (New York: Wiley, 1965), p.7; see also his A Framework for Political Analysis (Englewood Cliffs, N.J.: Prentice-Hall, 1965), and his "Alternative Strategies in Theoretical Research," in Easton, cd., Varieties of Political Theory (Englewood Cliffs, N.].: Prentice-Hall,

1966). 6 Adam Przeworski and Henry Teune, The Logic of Comparative Social Inquiry (New York: Wiley, 1970), p. 18. For a similar view of scientific theory, see Robert T. Holt and John E. Turner, eds., The Methodology of Comparative Research (New York: Free Press, 1970). 7Ernest Nagel, The Structure of Science (New York: Harcourt Brace Jovanovich,

1965), p. 4.

20

Behavioralism and the Three Faces of Power

Empiricism, Realism, and Power

sumption of events under "covering laws." This is sometimes called the deductive-nomological (D-N) view of science: laws are understood as conditional empirical statements whereby, given a general law and the statement of appropriate initial conditions, events can be deduced from the law. It thus follows that on this view explanation and prediction are seen as "symmetrical"-to be able to predict an event is to have explained it by bringing it under a covering law; and to be able to explain an event is simply to employ "retrospective predictions. "S

Following Rom Harre and Roy Bhaskar,' I label this understanding of science "empiricism" insofar as it takes the empirical world, the world of experienced occurrences, to be the object of scientific investigation, and it eschews any appeal to underlying causes and natural necessities as unscientific "metaphysics." It bears emphasis that this view extends beyond the positivistic view, now in utter disrepute, that theories are verifiable in experience and refer to unproblematic and unmediated observables.!O Many early behavioralists were positivists, but although positivism is a form of empiricism, the converse is not necessarily the case. The view I label empiricism hinges primarily on an ontology, or theory of reality, which can be traced back to David Hume-namely, that there is nothing but a flux of events whose only relationship is one of contingent conjunction. Hume contended that "all the objects of human reason or enquiry may naturally be divided into two kinds, to wit, Relations of ideas, and Matters of fact."" Relations of ideas, typified by mathematics 8See Carl Hempel, Aspects of Scientific Explanation (New York: Free Press, 1965), pp. 364-67. For a discussion of the bearing of this view on social expla~ nation, sec William H. Dray, ed., Philosophical Analysis and History (New York: Harper & Row, 1966). 9Cf. Rom Harre, Principles of Scientific Thinking (Oxford: Oxford University Press, 1970); and Roy Bhaskar, A Realist Theory of Science, 2d ed. (Atlantic Highlands, N.J.: Humanities Press, 1978). For a defense of this label, and the position it denotes, see Bas C. Van Fraassen, The Scientific Image (Oxford: Clarendon Press, 1980). lOFor a good discussion of the decline of positivism, see Harold I. Brown, Perception, Theory, and Commitment (Chicago: University of Chicago Press,

and logic, are characterized by analycity, or demonstrable certainty. Matters of fact, on the other hand, are elementary statements (to use more modern language) based upon "the present testimony of our senses." They are grounded in sense-data and map out that part of the word they represent. According to Hume, the only necessity is logical necessity. Matters of fact, by contrast, represent discrete and contingent experiences. He writes: "The contrary of every matter of fact is still possible; because it can never imply a contradiction, and is conceived by the mind with the same facility and distinctness, as if ever so conformable with reality. That the sun will not rise tomorrow is no less intelligible a proposition, and implies no more contradiction, than the affirmation that it win rise. We should in vain, therefore, try to demonstrate its falsehood. ,,12 This is the core of Hume's critique of the doctrine of causality and of the concept of natural necessity. Logic cannot provide us with the knowledge that the sun will rise tomorrow, or that objects will fall according to the law of gravity. Therefore, "cause and effect are discoverable, not by reason, but by experience." Causal knowledge is not necessary, but contingent: it "arises entirely from experience, when we find that any particular objects are constantly conjoined with each other."" This view of causality is widely accepted by philosophers and social scientists who are otherwise critics of Humean bedrock empiricism. Karl Popper, arguably the most important post-positivist philosopher of science, is most noted for his critique of Hume's sense-data epistemology. But he too subscribes to a Humean view of causality. Popper believes that a sense-data epistemology and a verificationist theory of meaning, two fundamental theses of logical positivism deriving from Hume, are dogmas. They are both logically unsound and misunderstand the real grounds of the objectivity of science. Popper argues that science cannot be grounded in experience, because of the unbridgeable gap between subjective experience and objective statements. Popper thus displaces the positivist insistence on verification with his own "criterion of demarcation." It is not that science, unlike metaphysics, is based on experience

1977). llDavid Hume, "Enquiry Concerning Human Understanding," in L. A. SelbyBigge, ed., Enquiries Concerning Human Understanding and the Principles of Morals (Oxford: Clarendon Press, 1975), p. 25.

21

"Ibid., pp. 25-26. !.lIbid., p. 27.

22

Empiricism, Realism, and Power

and is therefore the only meaningful form of knowledge. Science is demarcated from metaphysics by its method, and by the relationships that obtain between scientific propositions.!4 According to Popper, the distinctive feature of science is its deductive method.!S Because inference from singular statements to general theories is logically inadmissible, no theory is ever conclusively verifiable. But theories are "falsifiahle." They cannot be verified, but they can be refuted, by experience. Popper writes: "My proposal is based upon an asymmetry between verifiability and falsifiability, an asymmetry which results from the logical form of universal statements. For these are never derivable from singular statements, but can be contradicted by a singular statement. Consequently, it is possible to argue from the truth of singular statements to the falsity of universal statements."!' For Popper there are no ultimate statements, only falsifiable hypotheses. We can test these by deducing their consequences and seeing whether the predicted events occur. Of course, our tests themselves are not unmediated encounters with experience. They are describable only through scientific concepts that are themselves fallible and potentialiy subject to test. The "basic statements," which function as the jury in the trial of a theory, are thus not like the "protocol statements," which positivists believed could be verified through experience. They are distinguished by their logical form, not their origin.!7 They are, "in the material mode of speech, statements asserting that an observable event is occurring in an identifiable region of space and time."!8 In other words, they are singular statements, themselves fallible, which refer to specific events in time. Theories are thus tested empirically by deducing their consequences and comparing them with experience as articulated through basic statements. If the basic statements contradict the consequences that are deducible from the theory, then the prem14Karl R. Popper, The Logic of Scientific Discovery (New York: Harper Torch~ books, 1959), pp.34-39; see also his "The Demarcation between Science and Metaphysics," in Karl R. Popper, Conjectures and Refutations: The Growth of Scientific Knowledge (New York: Harper Torchbooks, 1965). lSPopper, Logic, p. 42. 16Ibid., p. 4 L l7Popper, Conjectures, pp. 50-51. 18Popper, Logic, p. 103.

Behavioralism and the Three Faces of Power

23

ise of the deduction must be deficient and the theory falsified. Of course, since basic statements are themselves fallible, there are no incorrigible foundations of truth. "From a logical point of view," says Popper, "the testing of a theory depends upon basic statements whose acceptance or rejection, in its turn, depends upon our decisions. Thus it is decisions which settle the fate of theories."!9 One major point of Popper's philosophy of science is to delegitimize those "pseudo-sciences" (Marxism and psychoanalysis) which always seem, according to Popper, to provide instances of verification, but cannot specify the conditions under which they would be falsified. Anything and everything can be explained, and any apparent inconsistencies explained away by means of ad hoc hypotheses (like the false consciousness of the proletariat or the resistance of the patient).20 Falsification is an alternative to verification which is quite suited to the task of rejecting these theories. Science thus becomes seen as a process of "conjectures and refutations." Scientists conjure np hypotheses that are "the free creations of our own minds." They then logically deduce the consequences of these hypotheses and test them. While theories cannot be conclusively proved, they can be decisively disproved, or refuted. Science is a method of trial and error, of holdly proposing theories, of trying our best to show that these are erroneous, and of accepting them tentatively and skeptically if our critical efforts are unsuccessful. 2! While most working scientists would think of science as an enterprise of constructing knowledge about substantive objects of knowledge, for Popper it is a process of destruction of conjectures, guided by a commitment not to substantive knowledge but to logical procedures. 22 The hallmark of the truly scientific theory, then, is that it is falsifiable, and that it states in advance the conditions under which it may be considered invalid. Theories must consist of deductive laws capable of predicting empirical events and of testing these predictions. Popper's philosophy represents an important departure from positivism in its decisive repudiation of any foundationalist theory "Ibid., p. 108. 20Popper, Conjectures, pp. 33-39. 21Ibid., p. 5l. 22See Jonathan Lieberson, "The 'Truth' of Karl Popper," New York Review of Books, November 1.8 and 25, 1982.

24

Empiricism, Realism, and Power

of knowledge. Popper has always insisted that "a science needs a point of view, and theoretical problems,"23 and has himself identified his philosophy with that of Kant insofar as both philosophers emphasize that knowledge is partly constituted by the human mind, which is not simply a passive recipient of impressions. 24 This dimension of Popper's thought has opened up the debate about "criticism and the growth of knowledge" and has stimulated the work of Thomas Kuhn, Imre Lakatos, and Paul Feyerabend on the decisionistic features of scientific practice. 25 This perspective is indispensable in social science and should serve to silence those positivistic researchers who insist upon "observabiliry" as a criterion of scientific meaning. And it has played an important role in the debate on power, enabling the critics of Dahl, for instance, to insist that no concept of power is without its presuppositions, its point of view. What is important for our purposes, however, is that Popper's philosophy, like that of Kant, is exclusively a criticism of positivist epistemology. Popper quite clearly accepts Hume's ontology and his account of causaliry. "To give a causal explanation of an event," writes Popper, "means to deduce a statement which describes it using as premises of the deduction one or more universal laws, together with certain singular statements, the initial conditions .... The initial conditions describe what is usually called the'cause' of the event in question. "26 On this view causaliry is nothing but the regular sequence of cause-and-effect events. For example, on Popper's view a causal explanation of the marching of a marine platoon would be: general law-whenever marine sergeants say "march" marine platoons march; initial condition-the marine sergeant said "march"; deduction-the platoon marched. Here the causal relation is nothing more than the regular conjunction of the sergeant's order and the platoon's marching, and the theoretical explanation nothing more than a generalization about this conjunction. (This is the ideal form of empiricist theories of power, and, I will suggest below, it is fundamentally deficient.) 23Popper, Logic, p. 106. 24Popper, Conjectures, pp. 189-93. 255ee Imre Lakatos and Alan Musgrave, eds., Criticism and the Growth of Knowledge (London: Cambridge University Press, 1970). 26Popper, Logic, pp. 59-60.

Behavioralism and the Three Faces of Power

25

Popper explicitly links himself with the Humean view, which denies that there are any necessary causal connections in nature. He quotes Wittgenstein paraphrasing Hume: "A necessity for one thing to happen because another has happened does not exist. There is only logical necessity."l7 Therefore "ultimate" explanationswhich attempt to understand the essential properties of things, and make "existential" statements about their natures instead of "conditional" statements about their predictable behaviors-are rejected by Popper as unscientific. He writes, describing science as "methodological nominalism": "Methodological essentialism, i.e., the theory that it is the aim of science to reveal essences and to describe them by means of definitions, can be better understood when contrasted with its opposite, methodological nominalism. Instead of aiming to find out what a thing really is, and at defining its true nature, methodological nominalism aims at describing how a thing behaves under various circumstances, and especially, whether there are any regularities in its behavior."28 There is in Popper's remark more than a trace of the kind of scorn Hume heaped upon medieval scholasticism and a suggestion that any talk of causal necessities and essential natures is akin to scholastic references to occult powers and cosmic teleology. This is an extremely constricted view of science, but regardless of its philosophical limitations, it has been taken up with a vengeance by contemporary political scientists. This has been particularly significant for the understanding of power. Jack Nagel, in his The Descriptive Analysis of Power accurately expresses the consensus when he writes that any talk of necessary causal connections smacks of "objectionable metaphysical implications.,,29 The First Face of Power: Behavioral Compliance The understanding of causality as constant conjunction, and ideal of scientific explanation as prediction, shaped a new and "Ibid., p. 436. 28Karl R. Popper, The Open Society and Its Enemies, vol. 1 (Princeton: Princeton University Press, 1966), p. 32; see also The Poverty of Historicism (New York: Harper Torchbooks, 1960), pp. 26-36. 29Jack Nagel, The Descriptive Analysis of Power (New Haven: Yale University Press, 1975), p. 36.

26

Empiricism, Realism, and Power

Behavioralism and the Three Faces of Power

rigorous effort to formalize the concept of power and to understand it as a causal relation between the behaviors of agents. Harold Laswell and Abraham Kaplan, in their influential Power and Society, WrOte that power "is a process in time, constituted by ex-

perientially localized and observable acts.',30 A number of articles expressed a similarly behavioral view. Thus James March wrote: "We can say that two individuals are in an influence relation if their behaviors are linked causally. "31 And Herbert Simon wrote: "For the assertion, 'A has power over B,' we can substitute the assertion, 'A's behavior causes B's behavior.'

,,32

This approach was taken up by Robert Dahl, one of the most important figures in the three faces of power debate. Dahl, in a series of methodological articles, asserted the need for a definition of power amenable to the kind of empirical research envisioned by behavioralism. Thus he wrote: "Power terms in modern social science refer to subsets of relations among social units such that the behavior of one or more units (the response units, R) depend in some circumstances on the behavior of other units (the controlling units, C)."33 Power, then, is an empirical regularity whereby the behavior of one agent causes the behavior of another. (We might also note that Dahl's language of stimulus and response has obvious affinities with B. F. Skinner's behaviorist psychology.) Dahl is very explicit about this. He continues: "For the assertion 'C has power over R' we can substitute the assertion Ie's behavior causes R's behavior,' ... the language of cause, like the language of power, is used to interpret situations in which there is a possibility that some event will intervene to change the order of other events."" It seems obvious that this notion of power rests on a Newtonian analogy. We are all naturally at rest or at constant velocity, until our move30Haroid Laswell and Abraham Kaplan, Power and Society (New Haven: Yale University Press, 1950), p. xiv. 31James G. March, "An Int.roduction to the Theory and Measurement of Influence," American Political Science Review 49 (1953),47. 3:!Herbert A. Simon, "Notes on the Observation and Measurement of Power," Journal of Politics 15 (1953), 5.ln the same vein, see Felix E. Oppenheim, "Degrees of Power and Freedom," American Political Science Review 54 (1960),437-46. 33Robert A. Dahl, "Power," in International Encyclopedia of the Social Sciences, vol. 12 (New York, Macmillan, 1968), p. 407. 34Ibid., p. 418.

27

ment is altered by an external force. Power is that force whereby social agents alter the behavior of other agents or, as Dahl puts it, get them to do what they would not otherwise do:" And true to his empiricism, Dahl insists that there are no necessary relationships between the behaviors of agents. He writes in his essay "Cause and Effect in the Study of Politics": "The only meaning that is strictly causal in the notion of power is one of regular sequence: that is, a regular sequence such that whenever A does something, what follows, or what probably follows, is an action by B."36 These remarks may sound unexceptionable, but their force must be emphasized. Dahl is making a positive statement about the meaning of the concept of power. But he is also making a negative statement, about what power does not mean. He is insisting that his concept of power does not smack of metaphysics, that it involves asserting nothing that is not empirically evident. This.causal view of power forms the basis of the entire three faces of power debate. All of the contestants agree that power is an empirical relation of cause and effect, and none of them conceives of power as involving any necessary connections, or what I will later call structural relationships. This is not to say that the reason for this is that subsequent participants in the debate consciously wished to endorse the Humean view of causality. It is, rather, simply that they failed to challenge it, most likely because they failed to recognize it-an interesting example of the power of a view that is neither asserted nor recognized as such, and that is sustained by virtue of its misrecognition as a simple fact of life. The controversy over power does not revolve around this major, Humean, premise. It revolves, rather, around the question "How do we identify those instances in which A gets B to do what B would not have done otherwise?" As Lukes points out, this question hinges on the question of a counterfactual-what would B have done otherwise? Dahl's answer to this is that B's revealed preferences indicate what B would have done. 37 Thus, "A has power over B" means that A's behavior regularly and predictably causes B to 35Robert A. Dahl, "The Concept of Power," Behavioral Science 12 (July 1957),

203-204. 36Robert A. Dahl, "Cause and Effect in the Study of I)olirics," in Daniel Lerner, ed., Cause and Effect (New York: Free Press, 1965), p. 94. 37Dahl, "Concept," pp. 203-204.

28

Empiricism, Realism, and Power

do something B does not want to do. This has been called the "first face of power" insofar as it involves instances of manifest conflict and compliance. And it has been called the "decisionist" view insofar as it limits itself to instances of actual decision making, or choice, in action. It is on the basis of this interpretation of the counterfactual that Dahl and his student, Nelson Polsby, insisted that any "scientific" claims about power must focus on instances of manifest conflict and demonstrate constant conjunctions of cause-and-effect behaviors. In so insisting, they employed their understanding of scientific method in order to delegitimize radical critics of American society who wrote about power and employed the language of power structure without referring to sequences of the sort just mentioned. Thus Polsby, in his influential and recently reissued Community Power and Political Theory, chastised what he called "categorialism," or categorical claims like" A has power over B," which fail to specify the empirical conditions, the causal behaviors, under which B can be predicted to act. (Note the similarity of this criticism to Popper's invidious distinction between methodological nominalism and methodological essentialism.) Polsby writes about the claim that there is a dominant class in a particular society: "For this latter statement to mean anything in a scientific sense, we must, according to the formal requirements postulated above, make reference to specific decisions in which particular outcomes are affected by members of the classes into which we divide the population, and secondly, we must state the conditions under which we can take it as demonstrated that the upper class does not have more power than the lower class. ,,38 Ascriptions of power, then, are conditional statements of regularity, falsifiable predictions about the "stimulii" of the powerful and the "responses" of the weak, which always refer to the "actual behavior" of the agents in question. 39 Similarly, Dahl, in his now classic "Critique of the Ruling Elite Model," criticized C. Wright Mills by asserting: "I do not see how anyone can suppose that he has established the dominance of a specific group in a community 38Nelson Polsby, Community Power and Political Theory (New Haven: Yale University Press, 1963; 2d ed., 1980), pp. 5-6. 39Ibid.) p. 121.

Behavioralism and the Three Faces of Power

29

or nation without basing his analysis on the careful examination of a series of concrete decisions."40 It is important to see what these sorts of criticisms accomplished, and doing so will clarify more precisely what this book sets out to challenge. On the one hand, some very sensible and plausible points are made regarding the importance of empirical evidence and the possibility of theoretical criticism. This is one way of reading a claim like Polsby's, that any assertion about power "must be put in such a form that in principle it is directly or indirectly subject to disproof by an appeal to evidence." On the other hand, the whip hand of science is being deployed in order to question the very meaning and reference of claims about power not conformable to Dahl's "decisionist" interpretation. Thus, Polsby's claim is not simply a sensible and innocent remark about evidence and criticism, but also a claim about the formal structure of properly scientific explanation, namely, that it must conform to some variant of the D-N model and must refer to empirical events and their uniformities. Part of the reason for the widespread influence of Polsby's critique is the conflation of these two very different sorts of claim. And part is, undoubtedly, due to the widespread acceptance of empiricism as second nature among social scientists. It is not Dahl's and Pols by's emphasis on the importance of empirical evidence, but their reliance on empiricism, on the Hume/Popper view of causality and scientific explanation, that is the problem with their view of power. And the dislodgement of empirical research from the confines of philosophical empiricism is one of the major tasks of the present analysis.

The Second Face of Power: Variations on the Same Theme This view, and its theoretical deployment, were immediately challenged by Bachrach and Baratz, who introduced the notion of a "second face of power."" Their criticism of the formulations of 4°Robert A. Dahl, "A Critique of the Ruling Elite Model," American Political Science Review 52 (1958), 463-64. 41Peter Bachrach and Morton Baran, "The Two Faces of Power," American Political Science Review 56 (1962), 942-52, and "Decisions and Nondecisions: An Analytic Framework/' American Political Science Review 57 (1963), 632-42.

30

Empiricism, Realism, and Power

Dahl rested on two basic points. The first, important but for our purposes secondary, was that Dahl and Pols by sometimes write in a naively positivistic vein, as if the analysis of power is self-evident, and the location of power unproblematic, simply a question of observation. Bachrach and Baratz insist that this is a mistake, that all science involves the making of judgments of significance deriving from the scientist's theoretical perspective. Here they draw upon post-positivist philosophy of science in order defend research into those aspects of political life that are covert and "nonobvious." The second point is that Dahl's formulation misses a crucial feature of what we ordinarily think of as power-the suppression of conflict. In criticizing Dahl's decisionist focus on actual conflict, they introduce the concept of the "nondecision," which they define as "a decision that results in suppression or thwarting of a latent or manifest challenge to the values or interests of the decisionmaker. ,,42 The point of this argument is to conceptualize not simply interaction, but limitations on interaction, as features of power. Their own formulation, however, is crucially ambiguous, opening them up to the charge that it is little different from Dahl's. They gesture at a structural formulation, conceiving power as implicated in institutionalized practices, but are ultimately unable to sustain this conceptualization. It is in gesturing toward a sttuctural view that they invoke E. E. Schattschneider's concept of the "mobilization of bias." Schattschneider wrote: "All forms of political organization have a bias in favor of some kinds of conflict and the suppression of others because organization is the mobilization of bias. Some issues are organized into politics while others are organized out. ,,43 Bachrach and Baratz argue that the way issues get "organized" in and out of social interaction has everything to do with power, and yet these features are ignored by an exclusive focus on actual behavioral conflict. They write: "Political systems and sub-systems develop a 'mobilization of bias; a set of predominant values, beliefs, rituals, These essays are reprinted in the authors' Power and Poverty (New York: Oxford University Press, 1970). 4~Bachrach and Baratz, Power and Poverty, pp. 43-44. 43E. E. Schattschneider, The Semi~Sovereign People (Hinsdale, Ill.: Dryden Press, 1960), p. 71.

Behavioralism and the Three Faces of Power

31

and institutional procedures ('rules of the game') that operate systematically and consistently to the benefit of certain groups and persons at the expense of others. Those who benefit are placed in a preferred position to defend and promote their vested interests. ,,44 This formulation comes dangerously close to postulating underlying structural relations determining behavior, risking the essentialism so scorned by properly trained, "scientific" theorists. Polsby writes: "The central problem is this: Even if we can show that a given status quo benefits some people disproportionately (as I think we can for any real world status quo), such a demonstration falls short of showing that the beneficiaries created the status quo, act in any meaningful way to maintain it, or could, in the future, act effectively to deter changes in it" [italics mine].45 Once again, the mark of scientificity is the examination of behavior; "a given status quo," in and of itself, holds no interest for the theorist of power. Bachrach and Baratz seem to find this sort of criticism compelling. And so Bachrach and Baratz sacrifice their interest in structure in the interest of "science." They explicitly state that power involves actual compliance, going so far as to assert that "it cannot be possessed," only exercised. 46 They admit that "although it is true that a nondecision is not visible to the naked eye, a latent issue is and so is the mobilization of bias ... the nondecision-making process is indeed subject to observation and analysis."47 And conceding to behavioralism, they hold that "although absence of conflict may be a non-event, a decision which results in prevention of conflict is very much an event-and an observable one to boot."4' In making these concessions, they confirm a point made by Geoffrey Debnam-that implicit in their formulation is an important distinction between power as nondecision and power as mobili44Bachrach and Baratz, Power and Poverty, pp. 43-44. 45polsby, Community Power, p. 208. See also Raymond Wolfinger, "Nondecisions and the Study of Local Politics," American Political Science Review 65 (1971), 1063-80 for a similar criticism. For an interesting critique of the positivism that Polsby and Wolfinger fall into, and a defense of inquiry into covert decisions of Bachrach and Baratz's sort, see Frederick Frey, "Comment: On Issues and Nonissues in the Study of Power." American Political Science Review 65 (1.971.), 1081-1101. 46Bachrach and Baratz, Power and Poverty, p. 19. 47Bachrach and Baratz, "Decisions and Nondecisions," p. 641. 48Bachrach and Baratz, Power and Poverty, p. 46.

r 32

Empiricism, Realism, and Power

zation of bias. The former refers to actual decisions, or actions, which involve behavioral causation, differing from the "first face" only insofar as it includes covert instances of suppression as well as overt instances of compliance. 49 The latter is an unexplicated and ultimately nonbehavioral phenomenon. Polsby's criticism is thus decisive: "How to study this second face of power? To what manifestations of social reality might the mobilization of bias refer? Are phenomena of this sort in principle amenable to empirical investigation?"SO Bachrach and Baratz never explicitly answer this question. Instead, they sacrifice their insight about the institutional basis of power to the scholarly "mobilization of bias" which we have labeled empiricism. Bachrach and Baratz's contribution to the debate was nonetheless important. Even if they did not get beyond a behavioral understanding of power, they opened up scholarly debate in two significant ways. First, in the course of making their argument about nondecisions, they introduced some interesting, if generally unrecognized, phenomenological questions into the debate regarding the rationality and intentions held by parties to a power relationship. 51 As I argue in Chapter 3, these questions regarding social ontology are crucial to thinking about social power. Dahl's language of stimulus and response is simply not adequate to the understanding of social power, and Bachrach and Baratz open up the possibility of making norms and ideologies central to thinking about power. Second, Bachrach and Baratz insist that all social research begins with presuppositions and is never simply a question of observation. Ironically, this insight was lost to them in their own discussion of the concept of power; if it had not been, they might have recognized the Humean presuppositions they shared with their antagonists. Nonetheless, however limited their discussion of power, this insight into the social production of knowledge played an important role in the critique of the dominant theory of power in American political science-plnralism. lt stimulated inquiries, like William Connolly's The Bias of Pluralism,52 into the implicit presuppositions 49Geoffrev Debnam "Nondecisions and Power: The Two Faces of Bachrach and Baratz,'; America~ Political Science Review 69 (September 1975), 889-999. 5opolsby, Community Power, p. 190. 51Bachrach and Baratz, Power and Poverty, pp. 17-38. 52William E. Connolly, Political Science and Ideology (New York: Atherton

Behavioralism and the Three Faces of Power

33

of pluralism as a substantive theory; and it thus opened up the possibility of inquiry into features of political life which, while not necessarily apparent to common sense, were nonetheless available to investigation:" Despite these contributions, however, Bachrach and Baratz did not get beyond the empiricist foundations of the discussion of power. In failing in this regard, they also failed to remove the stigma of "metaphysics" ftom the language of "power structure," failing as well to conceptualize the conditions of interaction as key dimensions of social power.

The Third Face of Power Steven Lukes, in his important Power: A Radical View picks up where Bachrach and Baratz left off. He calls their "two dimensional view" of power an advance over Dahl's "one dimensional" insistence that power involves only behavioral relations of compliance. He agrees that the study of power involves interpretative questions about which kinds of phenomena to study, but he also believes that Bachrach and Baratz have not probed deeply enough into the way collective action and social institutions shape power. He writes of their formulation: In the first place, its critique of behaviorism is too qualified, or, to put it another way, it is still committed to behaviorism-that is, to the study of overt, "actual behavior," of which "concrete decisions" in situations of conflict are seen as paradigmatic. In trying to assimilate all cases of exclusion of potential issues from the political agenda to the paradigm of a decision, it gives a misleading picture of the ways in which individuals and, above all, groups and institutions, succeed in excluding potential issues from the political process. Decisions are choices consciously and intentionally made by individuals between alternatives, whereas the bias of the system can be mobilized, recreated, and reinforced in ways that are neither consciously chosen nor the intended result of particular individuals' choices .... Moreover, the bias of the system is not simply sustained by a series of Press, 1967); William E. Connolly, ed., The Bias of Pluralism (New York: Athenon Press, 1969), 53See Frey, "Non decisions."

34

Behavioralism and the Three Faces of Power

Empiricism, Realism, and Power

individually chosen acts, but also, more importantly, by the socially structured and culturally patterned behavior of groups, and practices

of institutions which may indeed be manifested by individuals' inaction. 54

Lukes thus proposes that, if the concept of power is to take account of the way in which interaction is itself shaped and limited, then it cannot limit itself only to instances of behavioral compliance, as would the one- and two-dimensional views. He writes elsewhere: "Is not the supreme exercise of power to avert conflict and grievance by influencing, shaping, and determining the perceptions and preferences of others?"SS

Lukes submits that his view of power, and those of Dahl and Bachrach and Baratz, "can be seen as alternative interpretations and applications of one and the same underlying concept of power, according to which A exercises power over B when A affects B in a manner contrary to B's interests."" It is actually Lukes who makes the concept of interest central to the debate, but it is important to see how much similarity underlies his differences with his predecessors. Lukes agrees that power is a causal concept denoting behavioral regularities. To refer to our earlier example, he agrees that "A has power over B" means that A's behavior causes B to do something that B would not otherwise do. As Lukes puts it, "Any attribution of the exercise of power ... always implies a relevant counterfactual.,,57 In the cases of the first two faces of power, the counterfactual was provided by the existence of empirical conflict between the revealed preferences of A and B. Lukes's formulation differs from these views in insisting that preferences can themselves be the effect of the exercise of power. He thus insists that what B would otherwise do cannot be properly gauged by B's preferences, but rather by B's interest. Lukes then defines power as: "A exercises power over B when A affects B contrary to B's 54Steven Lukes, Power: A Radical View (London: Macmillan, 1974), pp. 2223. SSSteven Lukes, "Power and Authority," in Tom Sottamare and Robert Nisbet, eds., A History of Sociological Analysis (New York: Basic Books, 1978), p. 669. 56Lukes, Power, p. 27. "ibid., p. 41.

35

interest."·" The concept of power can thus refer to relations between A and B even in the absence of empirical conflict. Lukes contends that this view captures the essence of power as an empirical relatiou between A and B, and that the sole difference between this view and those articulated by his antagonists is that "those holding the three different views of power I have set out offer different interpretations of what are to count as interests and how they may be adversely affected."s, Lukes's own view is that the concept of interest, or what has been called "objective interest," has to do with what an agent would do under ideal democratic circumstances. It thus follows that if it can be argued plausibly that A affects B in a manner that limits B from doing what B would do under ideal conditions, then it properly can be said that A exercises power over

B.60

This notion of objective interest has been subjected to a great deal of criticism, some of which we will discuss below. But regardless of the merits of Lukes's notion of interest, the importance of the concept for him is grounded in his commitment to viewing power as an empirical regularity. Despite his differences with his antagonists, he explicitly states that he is merely interpreting a shared concept. Insofar as this is the case, Lukes's formulation (like that of Bachrach and Baratz) is ambiguous regarding those "socially structured and culturally patterned" dimensions of power he referred to in his critique of Bachrach and Baratz. A later essay, "Power and Structure," is intended to clarify this. S&lbid" pp. 22-25. 59Ibid., p. 27. 6°Ibid., pp. 34-35. This view of interest, as Lukes acknowledges, is drawn from William E. Connolly, "On 'Interests' in Politics," Politics and Society 2 (Summer 1972), 459-77. This conception owes much to the work of ]iirgen Habermas, particularly his Knowledge and Human Interests (Boston: Beacon, 1968). Lukes explicitly links himself to the idiom of critical theory in a later paper, "On the Relativity of Power," in S. C. Brown, ed., Philosophical D1sputes In the Social Sciences (Atlantic Highlands, N,j.: Humanities, 1979), p. 267. It is therefore curious that in a more recent paper Lukes rejects Habermas's (and his own earlier) transcendental conception of objective interest, opting instead for a Weberian subjectivism in many ways akin to Polsby's view. See Steven Lukes, "Of Gods and Demons: Habermas and Practical Reason," inJohn B. Thompson and David Held, eds., Habermas: Critical Debates (Cambridge, Mass.: MIT Press, 1982). This is an issue on which Lukes shows some confusion. For a critique, see Michael Bloch, Brian Heading, and Phillip Lawrence, "Power in Social Theory: A Non-Relative View," in Brown, Philosophical Disputes, pp. 243-60.

36

Empiricism, Realism, and Power

Here Lukes argues that there must be a synthesis of structural and empirical orientations, suggesting that there is a "dialectic of power and structure."61 Social structure limits action, and power, being an eventlike phenomenon, is discernible empirically. Power, he proposes, is an "agency" concept, not a "structural" one. Yet it "is held and exercised by agents (individual or collective) within systems and structural determinants. ,,62 This is a clarification of the relation between power and structure-social structure provides the limits within which power is exercised. But it also leaves unanswered the problem posed by Lukes's earlier discussion, in Power: A Radical View, of power in structural terms. [n other words, what is the nature of these structural determinants of power? How determining are they? If power is an agency concept rather than a structural one, and if it denotes behavioral regularities, then what precisely is the difference between Lukes's third face of power and the Bachrach and Baratz view he criticizes? [s it simply a focus on a different class of events, those involving the transgression of objective interest rather than simply compliance? If Lukes's view is different, his analytic bifurcation of power and structure doesn't go far in sho'wing us how. In short, Lukes is unable to articulate the structural nature of social power that, he rightly notices, is so important. This is because, while he has made an important contribution to the discussion of power by introducing the concepts of structure and agency into the debate, his discussion of these issues reveals some serious confusion. As we have seen, Lukes polarizes the concepts of structure and agency and insists that power is an agency category. In a recent defense of his views he writes: "Whether power is a form of structural determination or a concept tied to the notion of human agency (whether individual or collective) hinges on whether the alleged holder of power is correctly claimed to be able to act otherwise than he does."63 Lukes's emphasis on the question of human agency is meritorious, as is his more general concern with questions of social ontology-language, interpretation, being-which bear upon the nature of a specifically social 61Steven Lukes) "Power and Structure," in Steven Lukes, Essays in Social Theory (London: Macmillan, 1977). 62Lukes, "Power and Authority," p. 635; "On the Relativity," pp. 263-64. 6JLukes, "On the Relativity," p. 264.

Behavioralism and the Three Faces of Power

37

(versus natural) science. Lukes's reasoning, however, suffers from two difficulties. The first is that he fails to see, even though he cites Anthony Giddens's New Rules of Sociological Method (see Chapter 2), that the bifurcation of structure and agency is misplaced. In short, to think of human agency as structured is not to deny its significance or efficacy in any way.'4 Power can be structurally determined and exercised by intentional human agents who can always "do otherwise." This requires a view of determination different from the empiricist view, in which determination is understood as one thing regularly following another. And it requires a more elaborate conceptualization of structure and agency. As we will see in the next chapter, contemporary developments in realist philosophy of social science make such an understanding possible. But it is surprising that Lukes, who is one of the foremost social theorists in the contemporary Anglo-American tradition, simply takes for granted conventional understandings of structure and determination, especially given the literature he cites. If Lukes's first difficulty involves his polarization of structure and agency in favor of an agency approach, his second involves his inability to recognize 'who the real determinists arc. In "On the

Relativity of Power," Lukes associates proponents of a structural concept of power with the behaviorist psychology of B. F. Skinner. He claims that they both are interested in systems of "causal or statistical connections," and in this regard have a "cognitive interest in technical contro!''''' This claim is so remarkable because the intellectual cousins of B. F. Skinner are none other than the behavioralists with whom Lukes aligns himself. None of the contemporary structuralist theorists whom Lukes mentions-Louis Althusser and Nicos Poulantzas in particular-have ever advocated deductive-nomological or statistical theory. In fact they are all, in one way or another, committed to the kind of theoretical abstrac64Anthony Giddens, New Rules of Sociological Method: A Critique of Interpretative Sociologies (New York: Basic Books, 1976). Giddens has noted: "In representing structure as placing limitations or constraints on the activities of agents .. . Lukes tends to repeat the dualism of agency and structure that I have spoken of in earlier papers" (Anthony Giddens, Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis [Berkeley: University of California Press, 1979], p. 91). 65Lukes, "On the Relativity," pp. 266-68; Bloch et al., p. 251, also make this mistake.

40

Empiricism, Realism, and Power

CHAPTER 2

the masses in situations of actual conflict of revealed preferences; yet Dahl's view would prevent us from claiming this. (I do not mean, and do not believe, that Dahl would deny th!~, only that the logic of his articles about power would deny It.) And yet IS It necessary to argue about the "objective class interests" of the work. ers and peasants in order to say this? I should thmk not. Of course there is another possibility, one that appears startlingly commonsensical but that violates the basic premise of the three faces debate. That is, that the CPSU has power over the SOViet masses by virtue of the structure of Soviet society, in which pohtlcal power is monopolized by a single party. This claim about power, however, is clearly essentialist in Popper's sense, m that It IS mterested in the nature of Soviet society rather than m the search for behavioral uniformities. Such theoretical interests, therefore, require more than going beyond the three faces of power controvers~; they require the rejection of the empmClsm that IS the controversy s foundation. 69Dahl himself has on one occasion asserted ~hat his, v.i~w of power,~s ap~lic~ble only to democratic systems. He writes In detense ot hlS method: It. mlghl be objected that the test I have proposed wo.u1d. not ~ork in t.he most obvIous of all cases of ruling elites, namely in the totahtanan dictatorships. For the .control. of the elite over the expression of opinion is so great that OV~ftly there !S ~o d~sa­ greement; hence no cases on which to base a j~dg~ment anse. The objection IS a fair one. But we not concerned here with totalitanan syster.ns. We are ~oncerned with the application of the techniques of modern investi?atlon t? A~encan communities" (Dahl, "A Critique," p. 468). This is a peculIar quahfication to ake about a meta-theoretical position. One is tempted to ask Dahl on what basts he knows that elites have such control in totalitarian systems~ and that po~er ?oes not operate in similar ways in American communities. But this remark do~s m~lcate something I will suggest later in this book-rhat Dahl's theory of pluralism IS not a consequence of the application of his method.

n:

Realism and Social Scientific Theory It is widely claimed that we live in a post-positivist era. The positivist ideal of a perfectly formal and deductive science, tied indubitably to sensory perception through "correspondence rules," has been abandoned by the philosophical mainstream. Social scientists are well aware of this and therefore exhibit much less optimism about the possibility of such "ideal" explanations. It is common knowledge among social scientists that our investigations are shaped by our theoretical questions, that, as Popper has insisted, "a science needs a point of view, and questions."l The abandonment of positivism, and with it the naive (and selfrighteous) optimism of behavioral ism, has been an important achievement. Indeed, it has liberated social scientists from a frustrating and impossible ideal. But as is all too often the case in political struggle, the beneficiaries of hard-fought victory forget the battle, losing sight as well of the need to extend the struggle. So, while most contemporary social theorists are well aware of the demise of positivism, they all too often take the critique of positivism for granted. In doing so they risk failing to recognize the limits of their post-positivism. It is ironic that many of these theorists are the first to heap scorn on "another critique of positivism," mistakenly believing that its distorting legacy has been effectively dislodged. As we pointed out in the last chapter, empiricism is a doctrine lKari R. Popper, The Logic of Scientific Discovery (New York: Harper Torchbooks, 1959), p. 106.

41

42

Empiricism, Realism, and Power

more inclusive than positivism; and, further, many post-positivists are in fact empiricists. Many social theorists have not seen this, a fact that can be attributed to the strong hold Humean notions of causality and theoretical explanation have on us. They have thus not recognized the necessity of extending the criticism of positivism into a more general criticism of empiricism and its effects upon social scientific investigation. r In the last chapter we examined some of the consequences of empiricism for the conceptualization of power. In this chapter we will outline an increasingly accepted "realist" alternative to empiricism and suggest its relevance to social and political analysis. As we will see, realism builds upon the epistemological insight of post-positivism-that scientific analysis requires a theoretical framework. But realism goes further, subjecting to criticism the Humean understanding of causality and explanation as well.

Realism, Post-Positivism, and Scientific Theory Popper's rejection of foundationalism, and his belief that science is a process of critique rather than induction from experience, is an important source of post-positivist philosophy. There is, however, one crucial respect in which Popper fails to break with positivism: his commitment to a logical analysis of scientific practice. He has written on many occasions that "logic is the organon of criticism.'" His own critique of positivism was based in large part on his analysis of the logic (or rather illogic) of induction. And it is the logical certainty of Popper's deductive theory of science that provides its appeal for him. While Popper has always insisted that ultimately the falsification of a theory involves a scientific decision, a judgment, he has also insisted that in order for a theory to be properly scientific, it must be judged according to an invariable, formal decision rule. Of his method of falsification he writes: "Deduction, I contend, is not valid because we choose or decide to adopt its rules as a standard, or decree that they shall be accepted; 2Karl R. Popper, "The Logic of Social Science," in Theodor Adorno, ed., The Positivist Dispute in German Sociology (New York: Harper Torchbooks, 1971), p.98.

Realism and Social Scientific Theory

43

rather, it is valid because it adopts, and incorporates, the rules by which truth is transmitted from (logically stronger) premises to (logically weaker) conclusions, and by which falsity is retransmitted from conclusions to premises (this te-transmission of falsity makes formal logic the Organon of rational criticism-that is, of refutation)."3lt is this commitment to formal logic as the basis of scientific method, and the belief that to reject it is to ground science on arbitrary decrees and pure decisions-a kind of scientific Hobbesianism-which has generated the now famous controversy over scientific rationality and the growth of knowledge. Kuhn's concept of a "paradigm," Lakatos's concept of a "methodology of research programmes," and Feyerabend's "theoretical anarchism" all represent attempts to articulate the nature of scientific decision making.' While there are important differences between these authors, they share in common a "conventionalist" view of science-they all view science as a social practice, defined by a set of conventions, and undertaken by socially and historically situated practitioners. All of them have questioned what Lakatos has called Popper's "naive falsificationism "-the belief that a counterinstance constitutes sufficient grounds for the refutation of a scientific theory. In different ways all three have emphasized the fact that the practical-experimental and conceptual elements of scientific theory are always in the process of being worked out by scientists, and that no simple, logical canon can govern scientific decision making and determine the appropriateness or validity of a given theory. The adequacy of a theory is always determined by "Karl R. Popper, Conjectures and Refutations: The Growth of Scientific Knowledge (New York: Harper Torchbooks, 1965), p. 64. See also Popper's Objective Knowledge (Oxford: Clarendon Press, 1972). 4See Thomas Kuhn, The Structure of Scientific Revolutions (Chicago: University of Chicago Press, 1962), and The Essential Tension: Selected Studies in Scientific Tradition and Change (Chicago: University of Chicago Press, 1977); Imre Lakatos "Falsification and the Methodology of Scientific Research Programmes," in Imr~ Lakatos and Alan Musgrave, eds., Criticism and the Growth of Knowledge (London: Cambridge University Press, 1970). AIl of the essays in this volume are excellent and indispensable, particularly Kuhn's "Reflections on My Critics" and Paul Feyerabend's "Consolations for the Specialist"; and Paul Feyerabend, Against Method (London: Verso, 1978). On conventionalism, see Russell Keat and John Urry, Social Theory as Science (London: Routledge & Kegan Paul, 1975), pp. 4666. For a recent critical summary ofthis approach, see Richard]. Bernstein, Beyond Objectivism and Relativism (Philadelphia: University of Pennsylvania Press, 1983).

44

Empiricism, Realism, and Power

the considered judgments of members of scientific communities, by comparing it with other theories. In short, theories are underdetermined by the facts they explain. This conventionalist view of science captures many of the most characteristic features of scientific practice-the messiness of theoretical belief systems, which are not formally and deductively organized;' the reasoning by analogy and by iconic representation, which plays such a crucial role in scientific discovery,' and the controversy, and discontinuity, which tends to accompany scientific change.' Like Popper himself, however, his followers remain neoKantian with regard to the critique of Humean empiricism. While they have all emphasized the irreducibly human and social basis of theoretical knowledge, none of them has questioned the understanding of natural laws as empirical invariances and causality as constant conjunction. Like Kant, their critique has been exclusively epistemological-if Humean causality is to be possihle, then certain things must be true about the human production of knowledge of it. 8 Contemporary realism in the philosophy of science accepts the analysis of post-positivism-that science is a social piactice, that scientific language and concepts do not correspond with something immediately given in experience but are socially produced, and that their meaning is constituted through their use. Realism, as it is developed here, joins with post-positivism in rejecting any notion that science necessarily converges upon a twe representation of reality, and in insisting that scientific encounters with reality are necessarily mediated by language and interpretation.' On the realist 5S ee W. V. O. Quine, "Two Dogmas of Empiricism," in W. V. O. Quine, From

a Logical Point of View: Logico~Philosophical Essays (New York: Harper Torchbooks, 1953). 6See Mary Hesse, Models and Analogies in Science (Notre Dame, Ind.: University of Notre Dame Press, 1966), and "The Explanatory Function of Metaphor," in Mary Hesse, RetJolutions and Reconstructions in the Philosophy of Science (Bloomington: Indiana University Vress, 1980); Rom Harre, Principles of Scientific Thinking (Oxford: Oxford University Press, 1970); and Norwood Russell Hanson, Patterns of Discovery (Cambridge: Cambridge University Press, 1958). 7See John Krige, Science, Revolution, and Discontinuity (Atlantic Highlands, N.].: Humanities Press, 1980). 8S ee Roy Bhaskar, A Realist Theory of Science (Atlantic Highlands, N.J.: Humanities hess, 1978), p, 261. 9The version of realism developed here differs from that which offers a "realistic"

Realism and Social Scientific Theory

45

view, however, scientific practice is aimed at the development of models that explain real causal mechanisms.]O Realism is primarily an ontological doctrine, which rejects the Humean view of causality and causal explanation and insists that the world consists of a complex of mechanisms which causes the phenomena of expenence. In the realist view, things are not simply real in the weak, epistemological sense of existing independently of the knower. They are real in a strong, ontological sense, in that they are not necessarily isomorphic with experience at all. This point needs to be emphasized and clarified, because vety few philosophers, empiricists included, would deny that there are "real things" that exist.]] The positivist Moritz Schlick, in the classic and influential 1932 essay "Positivism and Realism," clearly articulates the prevailing empiricist view of "reality": "I must confess that I should repudiate and consider absurd any philosophical system that involved the assertion that clouds and stars, mountains and sea were unreal , that the chair by the wall ceased to exist whenever I turned my back.'>l2 He continues: "If one understood by positivism a view which denies the reality of bodies I must declare positivism to be simply absurd." In rejecting idealism, however,· Schlick defines "reality" as: "When we say of any object or event-which must i?terpr.etatio~ ~f theo~ies as corresponding to reality and converging upon truth. For thiS realistic realism, see Hilary Putnam, Meaning and the Moral Sciences (Boston: Routledge & Kegan Paul, 1978). Mary Hesse writes: "A feature of this type of realism is that it retains the logical presuppositions of empiricism, namely the accurate applicability of logic and an ideal scientific language to the world, ... In a wider perspective this realist problematic looks parochial and over-iotelleetu~l.ized, It has not only underestimated the challenge to empiricist presuppositions anstng from modern history and philosophy of science, but it also bypasses two other features of the general philosophical scene"-hermeneutics and the sociologv of knowledge ("Introduction," in Hesse, RelJo/utions, p. xiii). . IOSee Harre, Principles; Bhaskar, Realist Theory; Keat and Urry, Social Theory; Ted Benton, Philosophical Foundations of the Three Sociologies (London: Rout~edg~ & ~egan P.aul, 1977); Andrew Sayer, "Abstraction: A Realist lnterpretatIO~, R~dtcal Phtlosophy. (Summer 1981),6-15; Peter T. Manicas, "Reduction, EpIgenesIs, and Explanation," Journal for the Theory of Social Behaviour 13

(October 1983), 321-54. llSee Karl Popper's Realism and the Aim of Science (Totowa: Rowman and Littlefield, 1982). 12Moritz Schlick, "Positivism and Realism," in A. J. Ayer, ed., Logical Positivism (New York: Free Press, 1959), p. 97.

46

Empiricism, Realism, and Power

be designated by a description-that it is real this means that there exists a very definite connection between perception or other experiences, that under certain conditions certain data appear."13 In other words, of course there are things that are real; but to say that X is real is to say that X is an event we have experienced or can experience, or as Schlick puts it: "To be real always means to stand in a definite relationship to the given."" This view of the real, which Schlick labels "empirical realism," is the prevailing view. On it one doesn't deny reality; one insists that what is real is what is experienceable. Nelson Polsby, whose views of causality and explanation we have already encountered, is representative when he writes: "My idea of theory, a widely held but evidently not universally shared view, is that it exists primarily for the purpose of facilitating the storage of information about objects of empirical inquiry, and for producing knowledge ... about a world that has some existence independent and apart from the language in which scholars palaver" (italics mine)." Realism is based upon a more substantial view of what is entailed by the independence of the world from the observation of the world. Schlick considered it absurd that the chair by the wall ceased to exist whenever he turned his back. On the realist view, it is not enough to say that the chair exists whether I look at it or not; the fact that the chair exists, whether I look at it or not, means that there is no reason we should suppose that the ontological reality of the chair is nothing more than those (possible) experiences of it through which I can come to know it. For realism, the real is not simply the events we experience. However much we require experience in order to know about the world, and however much we can only experience events, the appearances of the world and the reality of the world should not be conflated. On this view science is precisely the use of reason to figure out what real mechanisms are causally responsible for the phenomena of experience. Realism thus rejects the understanding of causality as constant conjunction and scientific explanation as the prediction of empirical regularities. It defends the concept of "natural necessity," that nat13Ibid, "Ibid., p. 99. !SNelson Po!sby, Community Power and Political Theory, 2d ed. (New Haven: Yale University Press, 1980), p. 233.

1

Realism and Social Scientific Theory

47

ural (scientific) laws explain the properties and dispositions of things that are not reducible to their effects. For example, the physical properties of copper-malleability, fusibility, ductility, electrical conductivity-are not contingent effects caused by antecedent events; they are the enduring properties of copper as a metal, which can be accounted for by its atomic structure. On this view causality is understood as the actualization 01 the properties of real entities with "causal powers."" Scientists develop theories explaining the phenomena 01 experience (e.g., the fact that copper conducts electricity and string doesn't) via appeal to the structures which generate them. It is these structures, and the effects they tend to produce, which are the primary objects of scientific investigation.

Rom Harre and E. H. Madden, in their Causal Powers, provide an example. The earth possesses the power of gravitational attraction, although the barn won't collapse unless the center beam is removed. Physical theory, however, is not about whether or not the center beam will be removed ancl the barn collapse, an event that is purely contingent; it is about gravity and the gravitational properties of various masses. Similarly, fire has the power to burn human beings. 01 course, whether I get burned by fire IS purely contingent; I may never encounter a fire, or I may do so wearing an asbestos suit. But science tells us about the enduring properties 01 the world. As long as fire is fire and has the properties of fire (i.e., the production of heat), and as long as humans are humans and are so biochemically constituted, fire has the capacity to burn humans, whether or not it will do so on any particular occasions. The actualization of causal powers is contingent. But causal powers are naturally necessary-they are the way the things 01 the world tend to operate in virtue of what they are.l? On the realist view, the world is uot constituted such that it can be explained by subsuming eveuts under "covering laws" of the form "whenever A, then B." Rather, it is composed of a complex of what Harre has called "powerful particulars," or causal mechanisms, which operate in an unpredictable, but not undetermined, manner. Roy Bhaskar writes in his A Realist Theory of Science: 16Rom Harre and E. H.

tlefield, 1975). 17Ibid., chap. 1.

Madden, Causal Powers (Totowa: Rowman and Lit-

48

Empiricism, Realism, and Power

The world consists of things, not events .... On this conception of science it is concerned essentially with what kinds of things there are and with what they tend to do; it is only derivatively concerned with

predicting what is actually going to happen. It is only rarely, and normally under conditions which are artificially produced and con~

trolled, that scientists can do the latter. And when they do, its significance lies precisely in the light that it casts on the enduring natures and ways of acting of independently existing and trans factually active things. ls

This understanding of science does not eschew empirical evidence. But it construes this evidence as the means by which scientists come to explain underlying causes that are not apparent in experience. Bhaskar, in fact, argues that it is only such a view that . can make sense of the role of evidence and experiment in natural science. Bhaskar argues, through a "transcendental argument," that if science as we know it is to be possible, then the world must be characterized by natural necessity, and to put it negatively, empiricism must be wrong. Bhaskar's argument hinges on an analysis of experimentation in theoretical science. The scientist, making use of the skills, tools, and knowledge produced by the scientific community, closes the world through experiment so that knowledge of nature is possible. Under conditions of experimental closure, scientists isolate theoretically relevant phenomena and produce empirical regularities. These regularities are the evidence for the laws discovered by science. But, Bhaskar argues, if the universality of the laws is to be sustained, and the practice of experiment made intelligible, then the laws discovered and validated through experimental closure must be applicable outside their conditions of identification." In short, scientists identify natural laws by producing regularities through experiment. But clearly the practice of experiment would make no sense if these laws were considered real only under experimental conditions. Scientists produce empirical regularities in order to identify how things tend to behave even in the absence of experimental conditions. Experiments are necessary because causal mechanisms do not always produce their natural eflkBhaskar, Realist Theory, p. 51.

"Ibid., pp. 63-127.

Realism and Social Scientific Theory

49

fects in the open world outside of experimentally closed conditions, and when they do, they are often interfered with by other effects of other mechanisms. But experiments are intelligible only if we suppose that laws apply even in the absence of experimentally produced regularities, and if we view documenting the regularities not as the aim of science, but as the means of discovering and isolating the enduring causal mechanisms which produce them. Events in the world, on this view, are usually complexly caused by the conjunction of a variety of mechanisms.'o If we want to explain, for example, why my pipes broke on December 12, 1986, we must have recourse to certain explanatory theories that tell us about the properties of things like water (H 2 0 tends to freeze and expand at certain temperatures), copper piping, and so on. We must also have knowledge of more specific circumstances not built into the theories (e.g., the heater in my basement, which had prevented the pipes from freezing all winter long but broke on December 11). No general theory can have built into it enough specific information to enable us to deduce that the pipes will break in my basement, because the set of events (and decisions-why did I not replace that old heater last autumn?) pertinent to that occurrence are both idiographic and contingent. But given our theoretical knowledge and a more historical account filling in the specific circumstances in question, we can explain the breaking of the pipes (the temperature was below zero, water tends to free at below-zero temperatures, copper piping tends to expand but these pipes were defective, etc.). In the realist view theoretical science is about the properties of enduring mechanisms-like H 2 0-which are ascertained under experimentally produced circumstances. Given our knowledge of the various natural necessities in question, we can understand why, if the specific set of entirely unpredictable circumstances that occurred in my basement obtained, my pipes would break. But the circumstances themselves were not necessary. For realism the natural necessity of causal mechanisms is associated with the contin2°This example, and the more general analysis of explanation, is drawn from Peter T. Mani~as, "Structure and Explanation" (unpublished manuscript, Department of Philosophy, Queens College).

'>

50

Empiricism, Realism, and Power

gency of particular outcomes. Realism understands there to be determinants-the tendencies of real entities. But it is not deterministic. 21 It bears emphasizing that realism as a philosophy of science is not an esoteric or a priori doctrine, but is based on an analysis of the actual practice of science. For realism, the ontology of natural necessity is presupposed by what scientists do, in their classificatoty schemas, in their experimentation, and in their development of causal concepts. Stephen Toulmin has written of the scientist: "He begins with the conviction that things are not just happening (not even just happening regularly) but rather that some fixed set of laws or patterns or mechanisms accounts for Nature's following the course that it does, and that his understanding of these should guide his expectations. Furthermore, he has the beginnings of an idea what these laws and mechanisms are ... [and] he is looking for evidence which will show him how to trim and shape his ideas further. ... This is what makes 'phenomena' important for him."" Science is thus essentialist and metaphysical in Popper's (and Pols by's) invidious sense. It is interested in how things happen, primarily as a means of understanding why things happen, what mechanisms cause things to happen, and what their causal powers are. But it does not therefore presume any immutability or teleology about the world, nor does it presume that the world can be unproblematically and rationally perceived. Rather, it presumes that the natural world exists independently of human experience, that it has certain enduring properties, and that science, through the development and criticism of theoretical explanations, can come to have some (however imperfect) knowledge of it. No greater testimony can be ptovided on behalf of this view than that of Albert Einstein, who in a 1931 letter to Moritz Schlick wrote: "In general your presentation fails to correspond to my conceptual style insofar as I find your whole orientation so to speak much too positivistic. ... I tell you straight out: Physics is the attempt at the conceptual llBhaskar, Realist Theory, p. 106-107. See also Roy Bhaskar, "Emergence, Explanation, and Emancipation," in Paul Secord, ed., Conceptual Issues in the Human Sciences (Beverly Hills: Sage, 1981); and Michael Polanyi, "Emergence," in his The Tacit Dimension (New York: Doubleday/Anchor, 1967). :UStephen Toulmin, Foresight and Understanding (New York: Harper Torch~ books, 1961), p. 75.

"1·

I

Realism and Social Scientific Theory

51

construction of a model of the real world and its lawful structure . ... In short, I suffer under the (unsharp) separation of Reality of Experience and Reality of Being.... You will be astonished about the 'metaphysicist' Einstein. But every four- and two-legged animal is de facto this meta physicist. "23

Realism, Human Agency, and Social Practice In the realist view social science should be similarly concerned with the construction of models of the social world and its lawful structure. 24 Not behavioral regularities, but the enduring social relationships that structure interaction, would be the primary object of theoretical analysis. In proposing this, however, realists do not advocate a form of hyperdeterminism that reilies social structure. Rather, in criticizing empiricism in social study, they have drawn on those critics of social science who, harkening back to the nineteenth-century German Methodenstreit, maintain that social study requires a different mode of understanding (verstehen).25 Realism thus maintains that society is in important respects different from nature, and that empiricism has failed to take account of this. However, realism also insists that this stands as an indictment of social scientific empiricism, not the project of social science per se. The Humean view of explanation and ontology has never gone uncontested iu social theory. Giddens has usefully labeled the critics of the Humean view "interpretative sociologists."" They have argued that: (1) the social world, unlike the world of nature, must be understood as the skilled creation of active human subjects; (2) the constitution of this world as meaningful depends on language, regarded not simply as a system of signs but as a medium of practical and intentional activity; and (3) generating descriptions of 23Quoted in Gerald Holton, "Mach, Einstein, and tbe Search for Reality," Daedeius 67 (1968), 636-73. 24See my "Realism and Social Scientific Theory," Journal for the Theory of Social Behaviour 13 (Ocrober 1983), 301-8. 2SSee William Outhwaite, Understanding Social Life (London: Allen & Unwin, 1975), for a historical discussion of this approach. See also Richard Palmer, Hermeneutics (Evanston, Ill.: Northwestern University Press, 1969). 26 Anthony Giddens, New Rules of Sociological Method: A Critique of Interpretative Sociologies (New York: Basic Books, 1976).

52

Empiricism, Realism, and Power

Realism and Social Scientific Theory

53

social conduct depends on the hermeneutic task of "penetrating the frames of meaning which lay actors themselves draw upon in constituting and reconstituting the social world.'>27 Unlike postpositivist philosophy of natural science, "interpretative sociology" is an ontological doctrine-it is concerned with the nature of social being itself. And it insists that the differences between natural and human science should properly reflect real differences between nature and society. While natural science aims at the documentation of empirical regularities and the subsumption of events under predictive covering laws, human action is purposive and can only be understood by grasping the rules, norms, and concepts that are essentially (not contingently) related to action."' To understand the human world is to understand the purposes of human agents, and the conventions within which these are articulated. Perhaps the classic modern statement of this position is Peter Winch's The Idea of a Social Science. Here Winch insists that while natural scientists investigate causal regularities, social conduct is rule-governed and not causal. Social study, therefore, is radically unlike the study of nature. 29 This approach has attained a certain prominence among political theorists and has become something of a new counterorthodoxy in political science, particularly among those who consider themselves theorists but who reject the implications of behavioralist scientific theory. This explains the preoccupation of contemporary political theory with questions regarding the nature of language and subjectivity and the history of political concepts and speech.30 John Dunn summarizes this preoccupation: "The great strength of the hermeneutic approach is that it takes consciousness and action as the core subject matter of the human sciences and treats them

as essentially linguistic phenomena, the possibility of characterising which in language is constitutive of them in a sense in which it is not in the case of non-human nature."31 Perhaps the most influential proponent of this hermeneutic in political science is Charles Taylor. In a number of essays and books Taylor has argued that to understand society is to understand the norms and vocabularies constituting its practices. 32 In his now classic "Interpretation and the Sciences of Man," Taylor subjects to criticism the behavioralist project of "public opinion research," which is concerned with formulating cross-cultural empirical generalizations about the correlation of various opinions and behaviors.33 He argues that political actions presuppose a normative, "institutional" context. To meaningfully say that one has political opinions and performs political activities, we must first grasp the social understandings that constitute these. One can only vote if one participates in certain activities and possesses concepts about the meaning and rightness of these activities. Taylor writes:

"Ibid., p. 153. 28See Quentin Skinner, "Social Meaning and the Explanation of Social Action," in P. Laslett, W. G. Runciman, and Q. Skinner, cds., Philosophy, Politics, and Society, 4th ser. (Oxford: Blackwell, 1972). Skinner's analysis here draws heavily upon j. L. Austin's How to Do Things with Words (Cambridge, Mass.: Harvard University Press, 1962). 29Peter Winch, The Idea of a Social Science (London: Routledge & Kegan Paul, 1959). 30See my "After Empiricism: The Realist Alternative," in Terence Ball, ed., Idioms of Inquiry; Critique and Renewal in Political Theory (Albany: State University of New York Press, 1987).

. 31John Dunn, "Practicing History and Social Science on 'Realist' Assumptions," C. Hookway and P. Pettit, eds., Action and Interpretation: Studies in the Philosophy of the Social Sciences (Cambridge: Cambridge University Press, 1978), pp. 159-60. I should point out that by "realism" here Dunn is referring to the reality of an agent's intentionality, not to the reality of social structure. ..l2Charles Taylor, Philosophical Papers, 2 vols. (Cambridge: Cambridge University Press, 1985). JJCharles Taylor, "Interpretation and the Sciences of Man," Review of Metaphysics 25 (1971), 1-45. 34Charles Taylor, Hegel and Modern Society (Cambridge: Cambridge University Press, 1979), p. 88.

The point is that the objects of public experience-rite, festival, election, etc.-are not like facts of nature. For they are not entirely separable from the experience they give rise to. They are partly constituted by the ideas and interpretations which underly them. A given social practice, like voting in the ecclesia, or in a modern election, is what it is because of a set of commonly understood ideas and meanings, by which the depositing of stones in an urn, or the marking of bits of paper, COUnts as the making of a social decision. These ideas about what is going on are essential to define the institution. They are essential if there is to be voting here, and not some other activity which could be carried on by putting stones in urns. 34

In

54

Empiricism, Realism, and Power

Taylor thus contends that the explanation of human action must be in terms of the practical understandings, norms, and purposes of human agents. This kind of explanation, for Taylor, is radically different from the explanation of narure because, unlike nature, the object of human science is itself normatively constituted. Interpretative theory thus sees the social world as consisting of normatively governed practices rather than of contingently caused behaviors." This is a very important distinction, one that is lost, for instance, on Lukes, who mistakenly believes that the only way to make sense of human agency is through the lens of behavioralism. The rejection of empiricist views of human action does not mean that there are no behavioral regularities in society. It simply means that these regularities or patterns are not mechanistically or contingently connected. They are, rather, the ways in which activities are practiced, are purposively done according to certain social rules and self-understandings. Most things are done rather routinely in the course of everyday life and lack the quality of "events."" Moreover, they are not done simply as a matter of course, but in a variety of ways within normative limits. Thus, if we want to understand why children go to school in the morning, we would not formulate a law of the form "whenever it is 8:30 A.M., children go to school," or "whenever parents say so, children go to school." Rather, we would attempt to understand the social concepts and norms that make school going a meaningful activity routinely undertaken by children (e.g., respect for authority, belief in merit, obedience to parents). These norms are not, however, apodictic; they do not deterministically cause people to act in the manner in which a cue ball causes another billiard ball to move upon contact. Rather, they are both enabling and constraining, within limits; they are the materials that make possible what Sruart Hampshire has called "practical .USee Richard]. Bernstein, Praxis and Action: ContemlJOrary Philosophies of Human Activity (Philadelphia: University of Pennsylvania Press, 1976), and The Restructuring of Social al1d Political Theory (Philadelphia: University of Pennsylvania Press, 1978); and Anthony Giddens, Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis (Berkeley: University of California Press, 1979). J6See K. W. Kim, "The Limits of Behavioral Explanation in Politics," in Charles A. McCoy and John Playford, eds., Apolitical Politics: A Critique of Behavioralism (New York: Crowell, 1967), pp. 47-50.

Realism and Social Scientific Theory

55

reasoning".17_"the clarification of one's intentions and purposes," the "considered decisions" grounded in values and reasons, which are the stuff of human activity. Michael Oakeshott has observed: "All activity springs up within an already existing idiom of activity."" Practices can be thought of as idiomatic activities rather than as mechanistic behavioral regularities. The notion that society is constituted by the norms and concepts of its participants, and that interpretation and intentionality are distinctive properties of the human world, is a crucial one. As Anthony Giddens has pointed out, however, this insight is often wedded to philosophically idealist positions, which ignore: (1) the practical involvements of human life in material activity; (2) the causal conditions of human activity; (3) asymmetries of power and divisions of interest in society; and (4) the possibility of false consciousness and systematic mystification." To explain why I go to school requires recourse to my understandings and purposes (as well as my unconscious motives). But it also involves analyzing what are such things as schools, which are (both logically and historically) pregiven, which do not exist simply in virtue of my own beliefs about and reasons for attending them. and which mav in fact have properties that remain unacknowledged by ~~ ~-s'l participate in them. It should be obvious at this point that interpretative theory'S antiscientific approach is based upon the implicit acceptance of an empiricist view of science and causality. Winch is representative here. He does not question the Humean account of causality as constant conjunction; he questions whether, given the nature of science, society can be scientifically studied. 40 Interpretative theory has properly rejected the empiricist ideal of social study. But it has .17Stuart Hampshire, "Freedom and Explanation, or Seeing Double," in Alan Ryan, ed., The Idea of Freedom: Essays in Honor of Isaiah Berlin (Oxford: Oxford University Press, 1979). ~8Michael Oakeshott, Rationalism and Politics (New York: Free Press, 1962). wGiddens, New Rules, p. 155. On mystification, see Ernest Gellner, "Concepts and Society," and Alasdair MacIntyre, "The Jdea of a Social Science," and "Is Understanding Religion Compatible with Believing?" in Bryan Wilson, ed., Rationality (Oxford: Blackwell, 1970). 40See Roy Bhaskar, The Possibility of Naturalism: A Philosophical Critique of the Contemporary Human Sciences (Atlantic Highlands, N.J.: Humanities Press, 1979), pp. 169-95.

56

Empiricism, Realism, and Power

improperly rejected the project of a causal and explanatory social theory. A realist understanding of science is capable of appreciating the specificity of the human world and of agreeing, with Taylor, that social science must study the concepts and values of societies. It is capable, however, of recognizing the purposeful nature of human conduct and of seeing it as determined by causal mechanisms. These mechanisms, though, will be understood not as antecedent events but as the enduring, structural relations which enable and constrain human conduct. Realism and Social Structure The concept of social structure recently developed by realists is based on a categorical rejection of the bifurcation of structure and human agency. Anthony Giddens argues that there is a "duality of structure.,,41 He proposes that social structures are both the medium and the effect of human action. As such, they do not exist apart from the activities they govern and human agents' conceptions of these activities, but they are also a material condition of these activities. Giddens uses the analogy of language to illustrate this: there would be no language without speakers speaking, and yet language is at the same time the medium of speech. Language has structural properties that agents draw upon in order to perform communicative acts. Giddens draws this notion of the sui generis, pregiven nature of society, which provides the conditions for human conduct, from Emile Durkheim's classic formulation: When I fulfill my obligations as brother, husband, or citizen, when I execute my contracts, I perform duties which are defined, externally to myself and my acts, in law and in custom .... The system of signs I use to express my thought, the system of currency I employ to pay my debts, the instruments of credit I utilize in my commercial rela~ tions, the practices followed in my profession, etc,_ function independently of my own use of them .... Even when I free myself from these rules and violate them successfully I am always compelled to struggle with them. 42 41Giddens, New Rules, p. 12l. 42Emile Durkheim, Rules of Sociological Method (New Yor*: Free Press, 1966), pp.I-3.

Realism and Social Scientific Theory

57

Pace Durkheim, however, for Giddens society does not exert a "coercive force" on individuals such that they simply "internalize" its dictates. Society not only constrains, but it enables purposive human action. Those social roles available in society constitute the basis of human identity. Stripped of these roles, and of the activities associated with them, the individual would be nothing but a cipher. These roles, however, are also sustained by their continued performance by intentional human agents. Society is thus both the condition and the product of human action. The major point of this realist approach is that human agency has social-structural preconditions. On this view social structures consist of those relatively enduring social relations between agents in the performance of definite social practices. The family, for example, is a social structure, composed of enduring relations between its members (husbands/wives, parents/children) in the performance of specific social practices (child rearing, maintenance of the household, etc.).43 There are, of course, many different kinds of relations in society. The relation between two cyclists crossing paths is different from the more enduring, structural relation between a teacher and a student. The latter is what Bhaskar calls an internal relation. He writes: "A relation RAB may be defined as internal if and only if A would not be what it essentially is unless B is related to it in the way that it is.,,44 Social structures are enduring, internal relations. At the intersection of various socially structured relations are the chance encounters that play a not insignificant part in everyday life. Interaction, however, is social by virtue of the persistence of the ties binding action together. Thus, even the chance encounter between cyclists is made possible by more enduring relationships-citizenship, for example-which provide a common ground of interaction (the street) and a common set of signals, symbols, and expectations (glances, words) which constitute spatia-temporal proximity as interaction. 45 Bhaskar proposes that this realist view of social structure corresponds to a "transformationalist" view of human agency: human activity is understood as the purposive transformation of material 43See Peter T. Manicas, "On the Concept of Social Structure," Journal for the Theory of Social Behaviour 10 (1980), 65-82. 44Bhaskar, Possibility, p. 54. 45See Giddens, Central Problems, pp. 198-233.

58

Realism and Social Scientific Theory

Empiricism, Realism, and Power

(social) causes by human agents. People, in the course of their everyday lives, participate in social practices and relations, and draw upon their rules and resources, in order to achieve their own specific goals. Bhaskar summarizes this perspective: The conception I am proposing is that people, in their conscious activity, for the most part unconsciously reproduce (and occasionally

transform) the structures governing their substantive activities of production. Thus people do not marry to reproduce the nuclear family or work to sustain the capitalist economy. Yet it is nevertheless the unintended consequence (and inexorable result) of, as it is also a necessary condition for, their activity. Moreover, when social forms change, the explanation will not normally lie in the desires of agents to change them that way, though as a very important theoretical and

practical limit it may do

50.

46

Society, understood as an ensemble of enduring relations and practices, is thus reproduced and transformed through the intentional activity of concrete individuals and groups. Giddens argues that the concept of the structuring of social life is tied to the concept of structuration-the ongoing processes of action and interaction whereby structures are sustained or changed. "To study the structuration of a social system,H he writes, "is to study the ways in which that system, via the application of generative rules and resources, and in the context of unintended outcomes, is produced and reproduced in interaction.,,.7 E. P. Thompson has similarly argued that this process should be seen as a process of eventuation, where human beings, in their "unmastered practice,H make their own history, realizing certain historical possibilities. 48 Corresponding to this methodological distinction between social structure and intentional activity is a distinction between social scientific and historical explanation. Theoretical social science explains the general characteristics of enduring social relations that 46Bhaskar, Possibility, p. 44. 47 Anthony Giddens, A Contemporary Critique of Historical Materialism (Berkeley: University of California Press, 1981), p. 27. See also Giddens, "On the Theory of Structuration," in his Studies in Social and Political Theory (New York: Basic

Books, 1977). 48£. P. Thompson, The Poverty of Theory and Other Essays (New York: Monthly Review Press, 1978), pp. 45-50.

59

exist, as it were, "out of time" for particular agents. As Giddens puts it: "Structures exist in time-space only as moments recursively involved in the production and reproduction of social systems. Structures have only a 'virtual' existence.,,49 History tells us about the actions, events, and processes that actually occur in the concrete world, in specific times and places. Historical analysis is thus properly thought of as conjunctural-it involves the analysis of complex historical events that have many causes, including a set of structural conditions and the irreducible effectivity of individual and collective activity. This distinction, however, should not be reified. Social structures are real determinants of what happens in social life. But so are the intentions and purposes of concrete agents. Social structures exist "out of time" for specific agents, insofar as they are enduring, and usually unacknowledged, conditions of activity. But they are only relatively enduring. They govern action in time, they are reproduced in the process of interaction, and they are subject to historical transformation. They are, in short, historically specific. This has two important implications. The first is ontological-it must be made clear that when we undertake theoretical analvses of social structures (like capitalism or pluralist democracy) we' are performing what Giddens has called a "methodological epoche.";o That is, we are abstracting from the concrete processes of social interaction in specific times and places, in order to understand the general structural properties governing these interactions. In doing so, we recognize that these structures are no more or less than the ways in which historically situated human practices are done. Our "epoche" is in no way intended to undermine the theoretical or moral significance of human agency in process. The second implication is epistemological-as Mills argued, history is the shank of social science. 51 We cannot spin general theories of society from our heads. The analysis of social structure necessarily relies on historical and empirical knowledge, without which it is impossible to construct adequate theories. As Mills demonstrates in The Sociological 49Giddens, Contemporary Critique, p. 26. 5°Giddens, Central Problems, p. 80. 51C. Wright Mills, The Sociological Imagination (Oxford: Oxford University

Press, 1959), pp. 143-64.

Empiricism, Realism, and Power

60

Imagination, the refusal to properly ground theoretical explanation results in a hopelessly vacuous "grand theory." In avoiding this danger, however, we should not fall into its obverse, the pitfall of "abstract empiricism," in any of its forms. The historian Quentin Skinner, in properly objecting to deductivenomological accounts of historical events, once suggested that "the [political analyst's] primary aim should not be to explain, but only in the fullest detail to describe."" It is tempting to heed this suggestion and to believe that only concrete human activity is real. This too has its dangers though. On the realist view developed above, social structures are real determinants of the social world, despite their being nonempirical conditions of activity. If this is true, no matter how necessary historical and empirical knowledge is to theoretical analysis, it can never be sufficient. It is, in explanatory theory, the '''raw material" that requires, as Althusser insists,

theoretical labor in order to produce social explanations. 53 As Giddens points out, then, "in respect of sociology, the crucial task of nomological [i.e., general-theoretical] analysis is to be found in the explanation of the properties of structures."" This requires the work of theoretical abstraction. But Giddens also insists that "the identification of structures can in no sense be regarded as the only aim of sociological investigation. The instantiation of structures in the reproduction of social systems, as its medium and outcome, is the proper focus of sociological analysis."ss In other words, there is a complementarity between the two basic tasks of social and political theory, on the one hand analyzing the characteristics of specific social structures (e.g., capitalism), on the other investigating the way these structures are instantiated in the concrete events of the social world. 5lQuentin Skinner, "The Limits of Historical Explanations," Philosophy 41

Realism and Social Scientific Theory

61

Lukes, in one of his discussions of power as an "agency" categOty,56 invokes Stephen Cohen's Bukharin and the Bolshevik Revolution" as an example of a political analysis sensitive to the efficacy of agency and the relevance of choice to the question of power. But Cohen's book makes clear that this sensitivity hardly negates the importance of structural analysis. The Revolution of October 1917, for example, was a complex of events, of which Lenin's return to Petrograd by way of the Finland Station was crucial. An account of the Revolution that left out the story of such events and processes would be hopelessly inadequate (one of the great virtues of Cohen's account is that it teaches us how complex the process was and how many other events were necessary components of it). In this sense historical analysis must be narrative and concrete. If we are to understand the Revolution, however, we must also understand how Lenin's return (along with other occurrences, of course) could have precipitated a revolutionary upheaval. In order to do this, we must undertake an analysis of the structures and contradictions of Russian society, the international system and the balance of forces, the effects of World War I on these, and more. To understand all this is not simply to analyze the choices of individuals and groups; it is also to analyze the enduring relationships that characterized Russian society and the way they were instantiated in the Revolution as the medium of the revolutionary process and as the transformed product of the revolution. 58 It should be clear that on the view suggested here there are different levels of analysis and abstraction which must be combined in order to understand the concrete social world, and that these levels of analysis are not reducible to one another. 59 First, we must have theories of social structures-the family, the economy, the state (these examples are here purely illustrative; realism is agnostic about how to theoretically describe and interpret social structures).

(1966), 214 . .I3Louis Althusser, "From Capital to Marx's Philosophy," in Louis Althusser and Etienne Balibar, Reading Capital (London: New Left Books, 1970). For critical discussion, see Andrew Collier, "In Defense of Epistemology," in John Mepham and David-Hillel Ruben, eds., Issues in Marxist Philosophy, vol. 3 (Brighton, Sussex: Harvester Press, 1979), and Ted Beman, The Rise and Fall of Structural Marxism: Althusser and His Influence (New York: St. Martin's, 1984), esp. pp. 35-

51. 54Giddens, New Rules, p. 160. 55Giddens, Central Problems, p. 106.

56Steven Lukes, "Power and Structure," in his Essays in Social Theory (London: Macmillan, 1977). 57Stephen Cohen, Bukharin and the Bolshevik Revolution (Oxford: Oxford University Press, 1976). 58 1 draw here upon Peter T. Manicas, "Review Essay: States and Social Revo-

lutions," History and Theory 20 (1981), 204-18. 59See Althusser and Balibar, Reading Capital; and Nicos Poulamzas, Political Power and Social Classes (London: New Left Books, 1973), pp. 11-33.

62

Empiricism, Realism, and Power

Realism and Social Scientific Theory

63

These theories are abstract, and they make reference to hypothetical entities that, although determinants of empirical phenomena, are not themselves empirical. Capitalism, for example, is a historically specific social structure. But to speak of capitalism is not to speak of its instantiation in specific locales. It is to refer to its ptoperties as a real structure that always exists in complex relation to other structures in the concrete world, and that we can come to know through theoretical abstraction. Second, we must have theories of concrete societies, or what Althusser has called "social formations." At this level of abstraction we are not simply interested in understanding capitalism, but French (or British or American or Brazillian) capitalism. Here particular national and historical traditions, and the specific relationship between social structures in particular societies, come into play (e.g., the importance of the family enter. prise in French capitalism, or the connection between the authoritarian state and capitalism in Brazil). Finally, we must have interpretations of the historical moment-France in 1986, or New Haven in 1953. Atthis level of analysis, we apply theoretical knowledge of structures and social formations to specific events and processes, in order to provide theoretical explanations of historical conjunctures. Of course, at this level we must relax our methodological epoche; the ways in which participants define their situation, the networks of interaction, and the feints and jabs of strategic maneuvering are all relevant to the working out of events and the transformation of structures. But we must already have some structural theory if we are to understand how concrete agents in a specific time and place are shaped by and are shaping their society. Thus, for instance, Walter Dean Burnham's The Current Crisis in American Politics 60 provides an analysis of the current moment in American politics which integrates (1) a theoretical understanding of the way capitalism in general operates; (2) an analysis of the American social formation, with particular emphasis on the connection between class relationships, political party alignments, and electoral participation; and (3) a concrete analysis of the politics of Reaganism and its ascendancy. Burnham's analysis views

current political history as a process of structuration, viewing events with an eye toward their significance in the reshaping of political and economic relationships. How participants define their situations and organize to effect them are not mere epiphenomena. But their significance can only be grasped by understanding the structural context that is both the cause and the product of their activity.

60 Walter Dean Burnham, The Current Crisis in American Politics (Oxford: Oxford University Press, 1982).

61See James Farr, "Popper's Hermeneutics," Philosophy of the Social Sciences (1983).

Criticism and Social Scientific Knowledge The major point of this chapter is that realism is an ontological doctrine regarding the nature of causality and scientific explanation, and that this perspective makes intelligible the theoretical analysis of the properties of social structures. No such discussion could be complete, however, without some remarks on more properly epistemological issues. Without such a discussion, this book could easily fall prey to the criticism of empiricists of various stripes, who will express wonderment about the rationalistic implications of the realist understanding of theoretical abstraction and worry about the dogmatism that seems to be its consequence. One might note again Karl Popper's insistence that "a science needs a point of view, and theoretical problems," and note further his observation that "it is decisions which settle the fate of theories." As we noted, however, Popper never reconciles this view with

his own insistence on the scientific necessity of the deductivenomological method.61 The real burden of the realist argument here falls on the question of falsification. Popper's method, despite its inapplicability to the actual history of science, has always seemed to empirically minded researchers to be a clear means of scientific testing, of separating the scientific wheat from the chaff. Realism builds upon the arguments of Kuhn, Feyerabend, and Lakatos, namely that Popper's empirical falsification cannot be a sufficient basis of theory choice. Realism insists, further, that empirical prediction is not even a necessary condition of scientific knowledge. But if scientific explanation does not involve empirical prediction,

64

Empiricism, Realism, and Power

then how can theories be falsified? And if they cannot be falsified, then how are they properly scientific as opposed to, say, religious? The key word here is falsification. In the realist view scientific theories are most definitely susceptible to falsification, if we mean by this susceptible to criticism and refutation. Popper's great contribution to philosophy of science is his insistence that it is not "verificationism," but criticism, that is the hallmark of scientific practice. It is not this insight, but his own narrow interpretation of what scientific explanation and criticism entails, which realists reject. Before I proceed to a discussion of how theories can be constructed and evaluated, I should state that I do not believe there can be a blueprint for scientific activity, nor can there be an extrapractical formula for the evaluation of scientific theories. I do believe, however, that science is a rational endeavor in the specific sense that it involves the construction of arguments and the deployment of evidence, and the judgment of these on the basis of their cognitive adequacy. In this sense there are certain very general considerations relevant to the construction and evaluation of theories.

In the first section of this chapter we discussed Bhaskar's analysis of the nature of natural scientific experimentation. In his view scientists experimentally close off the world in order to isolate certain of its causal mechanisms. The empirical regularities that are experimentally produced are not, however, the aim of scientific investigation; they are only the means of isolating causes and of (predictively) testing theories. Now, it is a specific feature of society that experimental closure is a methodological impossibility. Experimental psychology is, of course, capable of designing small group experiments, through which some kinds of knowledge about the properties of human individuals is possible." But social science is primarily concerned with the production of knowledge about 62See Peter T. Manicas, "The Human Sciences: A Radical Separation of Psychology and the Social Sciences," in Paul F. Secord, ed., Explaining Human Behavior (Beverly Hills: Sage, 1.982); and Peter T. Manicas and Paul F. Secord, "Implications for Psychology of the New Philosophy of Science," American Psychologist 39 (April 1983) 399-413. On the episodic nature of social life and the impossibility of experimental closure, see Rom Harre and Paul F. Secord, The Explanation of Social Behaviour (Oxford: Blackwell, 1972).

Realism and Social Scientific Theory

65

social structures, those enduring relationships that are complexly instantiated in human practice. To "observe" these practices is necessarily to examine them in particular and highly localized conditions, and moreover in circumstances in which nO specific relationship is hermetically sealed. It is impossible to separate the typical behaviors of capitalists and workers in a factory, for instance, from the idiosyncratic behaviors of the people who occupy these roles; and it is further impossible to separate the effect of class relations on their conduct from, for example, the effect of gender or of citizenship. To even begin to do so would require a prior theoretical identification of these relationships. But even armed with our theoretical apparatus, we would still be unable to test our theories in the same way natural scientists can. Two important methodological implications follow from this. The first is that we cannot come to know about social structures by examining them in their "pure," isolated form. Second, we cannot decisively test social scientific theories as we can in natural science. Empirical prediction, which is only germane to natural science under situations of closure, cannot be a decisive epistemological feature of social scientific explanation. I will discuss each of these two points in turn. All science involves the development of theoretical abstractions in order to explain real objects presumed to exist independently of our theoretical inquiries. In natural science we can come to know about causal mechanisms by using our pregiven theories in order to set up experiments. Our observations, always through the lens of our theories, enable us to isolate causal mechanisms in their pure form. An experimental situation can thus enable us to experience the effects of a magnetic field in isolation from other causal processes operating outside of experimental conditions, and thus to develop and test relatively precise theories about magnetic fields per se. In social science we cannot have analogous access to social structures, and we thus cannot experience their specific effects in isolation from other causal processes. But as Giddens points out, this difference between nature and society-the fact that social structures are necessarily instantiated in the ongoing flow of interaction-provides us with another, specific means of identifying social structural relations. Social structures, unlike natural structures, are already partially constituted and differentiated by the

66

Empiricism, Realism, and Power

concepts participants have about them. Social science involves the development of theories that, at least in part, draw upon and transform the concepts of everyday life. This point decisively refutes naively positivistic versions of social science, with their talk of preconceptual "observation." We cannot simply observe interest groups, voting behavior, or political conflicts in unmediated form. As Taylor insists, we can only identify such phenomena because we possess concepts of them, concepts of what is involved in practicing such activities. The social scientific concept of "voting behavior," for instance, does not denote certain observable behaviors physicalistically or otherwise understood, but rather certain practices that are what they are partly by virtue of people's beliefs about them. Voting behavior is a social scientific concept parasitic upon, and derivative of, the notions people in liberal democratic societies have about electoral politics. This important epistemological point takes on added significance in light of the fact that not only how we know, but that which we know about, is determined at least partially by the concepts of everyday life. This is what Anthony Giddens has called the double hermeneutic of social science. 63 Like natural science, social inquiry involves the further development of an interpretative schema; but unlike natural science, it involves developing a hermeneutic framework about a realiry that is already preinterpreted by its participants. Social scientific theory construction always involves, then, the penetration and working up of everyday practical concepts, and the scientific differentiation of social phenomena is parasitic upon the lay identification of social practices. On the basis of these identifications (e.g., economy, money, marriage, voting, government) social scientists develop theoretical abstractions (e.g., unemployment equilibrium, surplus value, patriarchy, pluralist democracy). This process of theoretical abstraction involves not only a working up of lay concepts, but a dialectic, a dialogue with and a critique of other theories as well. In short, theory construction never takes place in a vacuum. Just like all other social practices, it involves the transformation of preexisting materials. If social scientific concepts have an epistemic origin, then, it lies in the 63Giddens, New Rules, pp. 148-54.

Realism and Social Scientific Theory

67

practical concepts of ordinary social practice, and not in some unmediated experience. 64 Once we have constructed a theory of a hypothetically real social structure (like "pluralist democracy" or "capitalist democracy," two different theoretical identifications of hypothetical characteristics of our political process), we are still left with the problem of how to evaluate and test it. Of course these are not, as Popper would have it, two separate and discrete processes, the spinning of a theory (which Popper calls the "free creations of our own minds") and then the testing of a theory after it is spun (which Popper insists is governed by strict canons of deductive logic).6'> To a very large extent, both processes necessarily occur simultaneously, the development of an explanation at once being a consideration of prior theoretical difficulties and possible future objections, the evaluation of a theory being at once the reworking and fortification of it. These processes, however, are methodologically distinct, as they must be. This is because the problem of testing and evaluation is really the problem of the grounds on which a theory is submitted to its opponents for evaluation and debate. While there may not be any simple, logical procedure governing such determinations, which are chronic features of scientific practice, realism insists that there is a necessary tension between the explanatory aspirations of theory and the elusiveness of reality. Theories may be underdetermined by facts, but they are not undetermined by them. The problem of theoretical criticism will thus not go away, and it cannot be made to so so by insisting that we cannot get outside of onr theories. For while it is true that we cannot get outside of some interpretative framework, this does not license any particular framework. In thinking about this problem, it is important to keep in mind that a scientific explanation is not the documentation of an empirical regularity, but a model of those enduring mechanisms that cause the phenomena of experience. Even natural scientific theories 64For criticisms of Schutz's "postulate of adequacy"-the view that social eXM planations must be adequate to the understandings of social agents-see Giddens, New Rules, pp. 27-33, and Bhaskar, Possibility, pp. 59-64. OSee Jonathan Lieberson, "The 'Truth' of Karl Popper," New York Review of Books, November 18 and 25, 1982.

68

Empiricism, Realism, and Power

are not predictions; it is just that we can test them, under certain conditions, through prediction. The problem for social science is thus not the ontological one of "to what do your theories refer if not to empirical events?" but the epistemological one of how to establish the validity of theories without falling into an invidious rationalism. This is a crucial problem, but it is a problem of testing, not of the meaning of claims about nonempirical entities. This problem is significantly lessened by the fact that we are usually confronted not with a single theoty, but with competing theories. As Lakatos points out, theoty choice is never the result of a simple "measuring" of a theory against reality. Rather, it involves deciding which of a number of tbeories has explanatoty power; we compare competing theories about reality rather than comparing a single theory to an unmediated reality. We can outline in the most general terms three criteria governing such determinations. 66 The first is exhaustiveness: a superior (or adequate) theory must be able to account for known phenomena and anticipate new phenomena by providing a fruitful research program. This, of course, involves the making of scientific judgments about the relevance of those things a theory can explain and about the future promise of its projected lines of inquiry. The second is independence: a theoretical explanation of a range of phenomena must postulate a model of a hypothesized mechanism that would account for those phenomena, the phenomena being explained not being a part of the explanation. Norwood Hanson points out that the hypothesis that the color and odor of chlorine are explained by the presence of atoms with this color and odor is debarred by this criterion. 67 The explanation of the phenomenon must appeal to a deeper level, to an underlying cause, and not simply reiterate the phenomenon itself. This does not mean that we must view the empirical qualities of chlorine as contingently related to an antecedent cause. It is quite proper to view them as the necessary effects of the causal properties of chlorine in virtue of its atomic structure. But the explanation must make independent reference to the atomic structure of chlorine such that it causes these empirical effects. 66For discussion, see Derek Sayer, Marx's Method: Ideology, Science, and Cri· tique in 'Capital' (Atlantic Highlands, N.].: Humanities Press, 1979), pp. 115-17. 67Hanson, pp. 86-88,

Realism and Social Scientific Theory

69

The third criterion is consistency: a theory must be internally and logically consistent. This should not be understood as the unrealistic requirement that theories be organized as rigidly deductive systems of propositions, and that they be falsifiable in Popper's sense. As W. V. O. Quine points out, theoretical systems involve much more complicated networks of beliefs: The totality of our so-called knowledge or beliefs, from the most casual matters of geography and history to the profoundist laws of atomic physics or even of pure mathematics and logic, is a manmade fabric which impinges on experience only along the edges ...

[it] is like a field of force whose boundary conditions are experience. A conflict with experience at the periphery occasions readjustments in the interior of the field. Truth values have to be redistributed over some of our statements. Reevaluation of some statements entails reevaluation of others, because of their logical interconnections-the logical laws being in turn simply further statements of the system, certain further elements of the field. Having reevaluated one statemeilt we must reevaluate some others .... But the total field is so underdetermined by its boundary conditions, experience, that there is much latitude of choice as to what statements to reevaluate in the light of any single contrary experience. No particular experiences are linked with any particular statements in the interior of the field, except indirectly through considerations of equilibrium affecting the field as a whole. 68

The point is that logic imposes constraints upon theories by compelling redefinitions and readjustments of our beliefs. It is oftensupposed, for instance, that Marx's theory of the tendency of the rate of profit to fall has been conclusively disconflrmed by empirical evidence, calling into question the entire edifice of his economic theory." As Quine's comment would suggest, however,this belief about the rate of profit is connected to many other beliefs-about the class relations of capitalism, about the abstract understanding of the operation of markets, and more. No amount of empirical evidence about the rate of profit can, therefore, conclusively refute Marxian economics. But the constraint of logical consistency re6SQuine, pp. 42-43. 69For a discussion of these issues, see Ian Steedman et al., The Value Controversy (London: Verso, 1981).

70

Empiricism, Realism, and Power

quires that this theory equilibrate itself in the face of the evidence, using its basic concepts to invoke causes-counter tendencies, contingent empirical occurrences-to restore its integrity. This does not mean that any old ad hoc explanation will do, as, for instance, Popper fears. Basic terms of the theory must be mobilized to preserve "equilibrium." And in so doing, theorists will also be simultaneously making judgments regarding the satisfaction of the other two criteria, exhaustiveness and independence. It maybe, for instance, that another theory can better explain empirical trends regarding profit rates, accounting for the phenomena explained by Marxism and explaining other things besides. It does mean that theoretical practice, and the dialectic between competing theories, is a protracted process of contestation and reasoned judgment in which there can be no Archimedean point of scientific cerrainty .70 I would suggest, then, that while there are certain conditions of argumentative adequacy that any social scientific theory must satisfy, there are no extra-scientific considerations that can simpliciter ground theoretical decisions and judgments. Substantive questions will ultimately ground theoretical commitments. This troubles those empiricists who remain wedded to an objectivist view of

knowledge as correspondence with something given in experience, and of epistemology as the general, logical requirements of all valid knowledge. This is, however, the only possible way to do science, and the way it is done, even by social scientists who believe otherwise. Theoretical disputes in "democratic theory" revolve not around the falsification of regularity statements (of which there are few, if any), but around the ability of different theories to account for a range of phenomena-social inequalities, political conflict, and more.71 Dispntes in contemporary theory of the state involve arguments about what the state is and how various theories can account for different forms of state, wars, and revolutions. n If the 70See Peter T. Manicas and Alan Rosenberg, "Naturalism, Epistemological In¥ dividualism, and the 'Strong Programme' in the Sociology of Knowledge," Journal

for the Theory of Social Behaviour 15 (March 1985),76-101. 71See David Greenstone, "Group Theories," in Fred I. Greenstein and Nelson Polsby, cds., Handbook of Political Science (Reading, Mass.: AddisonHWesley, 1975), pp. 243-318. 72See Martin Carnoy, The State and Political Theory (Princeton: Princeton University Press, 1984).

Realism and Social Scientific Theory

71

remainder of this book is correct, and there is no theory of power conforming to the canons of empiricism and Marxism represents a realist theory of social power, then contemporary disagreements about the distribution of power are similarly to be resolved by arguments possessing not apodictic, but simply persuasive force. For realism, then, while empirical facts and evidence are absolutely indispensable components of theoretical inquiry, it is not facts, but judgments about the way theories explain a range of facts, which alone can determine the adequacy of theories. This epistemological perspective is not peculiar to realism; as we have seen, it is accepted by a number of important post-positivist philosophers. What distinguishes realism is its insistence that the aim of social science is to explain causal mechanisms that are not reducible to their empirical effects, which are, so to speak, real but not empirical. Realism thus claims that the adequacy of theories is grounded in judgments regarding their ability to explain a world that exists independently of our theoretical inquiries, and that defies the regulative ideal of deductive-nomological explanation. As Paul Feyerabend points out, any apodictic, extra-scientific criterion of theoretical choice can only be a conservative dognlait can never provide sufficient reasons for choosing, only establish formal canons requiring institutional sanctions for their force. 73 It would not be too cynical to suspect that this is the way empiricism has functioned as a hegemonic method in contemporary social science. As we shall see in the next chapter, the contestation of power is not the same thing as its negation, and so despite the growing skepticism about social scientific empiricism, it would not be wrong to consider it the dominant perspective in social science. In proposing its abandonment, I do not mean to imply that one intellectual straitjacket be replaced by another. The great virtue of realism is that, in recognizing the necessary limits of any methodological approach, it can shift social inquiry from a concern with specIOus canons of scientificiry to a concern with substantive explanation. 73Feyerabend, pp. 196-99.

The Concept of Power Revisited CHAPTER 3

The Concept of Power Revisited We are now in a better position to appreciate the basic weakness of the three faces of power debate and to reformulate the concept of power. The behavioralist foundations of the debate constrained its participants from conceiving of power as anything more than a behavioral regularity and prevented them from seeing it as an enduring capacity. To do so, of course, risks presupposing what Jack Nagel has called "objectionable metaphysical implications." But as we have seen, on another view of science-realism-such presuppositions about the enduring natures of causal mechanisms are the essence of actual scientific practice. It is only at great cost that the discussants of power have eschewed such premises. In this chapter I will outline a realist analysis of the concept of power and its role in social and political inquiry. I will argue that a power is an enduring capacity to act, which mayor may not be exercised on any particular occasion. And I will suggest that social power is implicated in social structure and is a necessary feature of human agency. In advancing this argument, I will challenge the behavioralist focus of the three faces debate, as well as its view that the concept of "power structure" is a metaphysical construct at odds with properly social scientific analysis. A realist view of power, we shall see, allows discussion of a number of features of power ignored in the three faces debate-domination, reciprocity, and ideology.

73

Power as Capacity Possibly the most glaring deficiency of the three faces debate is its failure to discuss what power is, instead stipulating in the most general way what "power terms" in social science have traditionally meallt. Like all words, power can be used in a variety of contexts. As Hannah Pitkin has pointed out, empiricist theorists of power have flagrantly abused language by their inattention to linguistic complexities and to questions of meaning. ' Witness, for example, Nagel's observation: "Words, as Humpty Dumpty observed, can mean anything we choose them to mean. Why bother to dispute definitions? I do so precisely because definitions are merely arbitrary, whereas hypotheses are potentially subject to agreement producing tests. Therefore, the most useful definitions are those which direct efforts to empirical research. ,,' This was, as we have seen, the attitude of the behavioralist innovators regarding the concept of power-that the concept should acquire a formal definition amenable to their understanding of scientific explanation and falsification. This effort was a striking failure in its own terms. If the most useful definitions are those that direct efforts toward empirical research, then the three faces of power debate must be judged useless, as it has generated a dearth of research actually conforming to the methods it prescribed. This attitude is frivolously mistaken, however, and in a way that sheds light on the theoretical sterility of the debate. Words can be used only in the context of their previons usage. The empiricist theorists of power have confined themselves to one particular locution, "power over," corresponding to their belief that a proper social science is a science of behavioral regularities. What is crucial is that they have all quite carelessly failed to provide a real definition of power, substituting instead an operational definition of the form "A has power over B means that whenever A does Y, B does Z." Power, a "potential" word, becomes redefined to describe contingent regularities. As Pitkin has observed, however, such an operationalist attitude toward language is selfIHannah Feneichel Pitkin, Wittgenstein and Justice (Berkeley: University of California Press, 1972), pp. 264-86. 2Jack Nagel, The Descriptive Analysis of Power (New Haven: Yale University Press, 1975), p.175.

72

74

Empiricism, Realism, and Power

defeating: "Operational definitions ultimately are useful if they come close to the real definition; if our operational definition of 'power' is not related to the meaning of 'power' then the results of any study we conduct with it will not yield information about power. "3 It is thus not surprising that the three faces of power controversy remains unresolved, and that the parameters of the debate prevent us from saying many things about power we should want to say. The word power derives from the Latin potere, meaning "to be able." It is generally used to designate a property, capacity, or the wherewithal to effect things.' The attribution of properties or capacities is a common feature of everyday life (e.g., "that fire is hot," "this car is fast," "your friend is smart"). This does not mean that ordinary ascriptions constitute valid scientific explanations, but it .does indicate the congruence of the ordinary sense of the term with the arguments we are developing. According to the realist philosophy of science outlined above, powers are a central subject matter of natutal science. Harre writes : "To ascribe a power to a thing or material is to say something about what it will or can do ... in virtue of its intrinsic nature.,,5 Power is, then, a causal concept;

but it is causal in the realist sense, as it refers to necessary properties, not contingent effects. To use an earlier example, to assert that conductivity is a power of copper is to claim that copper possesses the enduring capacity to conduct electricity, in virtue of its intrinsic nature, in this case its atomic structure.

Social science should be similarly concerned with the ascription of powers to social agents, and with the explanatory reference of these powers to agents' intrinsic natures. By the intrinsic natures of social agents I mean not their unique characteristics as individ3Pitkin, p. 275. On real definitions, sec Petcr T. Manicas and Arthur N. Kruger, Logic: The Essentials (New York, McGraw-Hill, 1976), pp. 34-38. 4See "Power," Oxford English Dictionary (Oxford: Clarendon Press, 1933), p. 1213. See also Pitkin, pp.274-79; Quentin Gibson, "Power," Philosophy of the Social Sciences (1971), 101.-12; and Stuart Clegg, "Power, Theorizing, and Nihilism," Theory and Society 3 (Spring 1976), 65-85, In what follows I am building on Terence Ball's "Power, Causation, and Explanation," Polity 8 (Winter 1975), 189-214, and his "Models of Power: Past and Present," Journal of the History of the Behavioral Sciences (July 1975),211-22. sRom Harre, "Powers," British Journal of the Philosophy of Science 21 (1970), 85.

The Concept of Power Revisited

75

uals, but their social identities as participants in enduring, socially structured relationships. Theories of power, then, should be conceived as interpretative models, developed by social scientists as submitted to the rigors of critical consideration, about the social structures which shape human action and distribute the capacities to act among social agents. And at the level of epistemology, to speak of the social structures that account for power is no different from speaking of the atomic structure that accounts for conductivity. Both sorts of claims are equally fallible, equally subject to theoretical and empirical criticism, and equally about underlying and non observable causal mechanisms (the important differences between nature and society we have discussed above notwithstanding).

Power and Agency As we will recall, Steven Lukes identifies the empiricist view of power with an "agency approach," insisting that to think of power in structural terms is to reify human activity. It is true that on the empiricist view power is exhausted in interaction, as behavioral regularity, and thus the empiricist formulation recognized no distinction between the possession and the exercise of power. It would thus seem to have the merit of placing agency at the center of social analysis. A few clarifications of the relationship between power and agency, however, will show that Lukes's claim is highly deceptive. First, on the realist view of power as capacity, power is not a specific kind of practice; it is implicated in all social practice, as a logically necessary feature of activity. For empiricism the exercise of power is a specific kind of act, one in which an actor gets another actor to do something that would not otherwise be done. This is a restrictive view of power, which fails to see that all human action entails the capacity to perform the activity in question. Power is implicated no more or less in my typing of this page, or my lecturing to my class, than it is in my getting my employer to comply with my demands for a raise. Giddens writes: "The relation between the concepts of action and power, on the level of strategic conduct, can be set out as follows. Action involves intervention in events in

76

Empiricism, Realism, and Power

the world, thus producing definite outcomes .... Power as transformative capacity can then be taken to refer to agents' capabilities of reaching such outcomes. ,,6 Not only does the realist view place agency at the center of power, as the exercise of power; it also places power at the center of agency, as a properry of huma~ agents that makes their activiry possible. Second, not only does the empiricist view fail to recognize that power is a necessary feature of action, it also misconceives the nature of action. We have seen that, for empiricism, the aim of social science is the discovery of behavioral regularities and the formulation of predictive generalizations. As interpretative critics have observed, this explanatory ideal fails to countenance the idiographic dimension of social life, which, constituted by the specific meanings and purposes of agents, defies predictive generalization. In short, em piricist theorists of power have failed to make the distinction be tween behavior and practice we noted in Chapter 2. Robert Dahl's language of stimulus and response is only the extreme form this inattention to the specificity of human agency has taken. On the realist view, power is implicated in all social life as the capacity to perform intentional activities and to engage in n~f­ inatively constituted practices. To speak of the power of an agent is thus to speak of the things an agent can do, where doing is understood not as behaving in response to an antecedent cause, but as performing a practical activity according to certain understandings and reasons. These understandings and teleologies, or ends toward which activity is directed, are not incidental but essential dimensions of action and of power. Thus, instead of asking about the regularity with which teachers gets their students to do homework, an interest in practice would direct our attention to the practice of education itself, to those norms defining the activities of its participants and those capacities entailed by these activities. As we saw in Chapter 2, this view of activity as rule-governed is tied to a view of activiry as rule-interpreting. There is thus an inherent indeterminacy in social life, the future being caused in part by the definitions of reality of social agents. This is different from Anthony Giddens, Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis (Berkeley; University of California Press, 1979), p.88. 6

The Concept of Power Revisited

77

saying, as for instance Dahl does, that theories of power are necessarily incomplete, because there are "an indefinite number of critical links in the chain of causation and therefore an indefinite number of 'faces' of power.,,7 Dahl's point is an epistemological one-the world is complex, and knowledge of it hard to come by. Our point is an ontological one-the ~()rld is open, and it is crucially determined by the purposive choices of social agents. Causaliry is thus not conceived-;'s chain in ~hi~h particular behaviors are but "critical links." It is thought of as an intricate web of determinants, structural and purposive, which complexly cause social interaction. Peter Winch has similarly noted the failure of social scientific empiricism to appreciate the hermeneutic dimension of power:

'a

An event's character as an act of obedience is intrinsic to it in a way which is not true of an event's character as a dap of thunder, and this is in general true of human acts as opposed to natural events. .. . There existed electrical storms and thunder long before there were human beings to form concepts of them .... But it does not make much sense to suppose that human beings might have been issuing commands and obeying them before they came to form the concept of command and obedience. For their performance of such acts is itself the chief manifestation of their possession of those concepts. An act of obedience itself contains, as an essential element, a recognition of what went before it as an order.s

Winch rightly insists that the exercise of power presupposes a normative and institutional context. A command presumes some mode of mutual understanding, and obedience some "uptake" of the command. My instructions to my class to read Machiavelli, and their obeying of these instructions, presuppose a set of shared concepts (teacher/student, homework/grades) and norms (regarding school, the value of learning, the value of good grades). The outcome of classroom interaction depends on these concepts, and how they are interpreted by teachers and students. (We will return to "

,

7Robert A. Dahl, Modern Political Analysis, 3d ed. (Englewood Cliffs, N.J.: Prentice-Hall, 1976), pp. 39-40. 8Peter Winch, "The Idea of a Social Science," in Bryan Wilson, ed., Rationality (Oxford, Blackwell, 1970), pp. 9-10.

78

Empiricism, Realism, and Power

The Concept of Power Revisited

79

, the question of the negotiation of power below.) But while the exercise of power certainly presupposes some social concepts, this does not necessarily mean that it presupposes the concept of the exercise of power. In fact, although social relationships of power involve some element of mutual recognition on the part of their participants, most such relationships do not involve the full recognition of the character of the relationship itself. Most relationships of power, in short, are not recognized as such. The subordination of slaves, for example, certainly involves some concept on their part of "'master" and some notion of "obedience." But it also typically involves other beliefs-about God, natural inferiority-which obscure and mystify this relationship. As we saw in Chapter 2, the great strength of interpretative theory is its insistence upon the hermeneutic character of social life. We also saw, however, that social life is only partly constituted by the concepts of its participants. It is also constituted by a set of enduring structural relationships that are likely opaque to their participants. Ernest Gellner, for instance, discusses the example of igurramen, the privileged elite of Moroccan Berber tribal society, who are believed to possess baraka, a magical power of plenitude bestowed by God. While the elite is in fact recruited through a complicated process of patronage and election, the concepts of baraka and igurramen are constitutive of this process, providing it with both meaning and justification; as Gellner points out, "There is here a crucial divergence between concept and reality, a divergence which moreover is quite essential for the working of the social system." He thus argues that a purely hermeneutic social analysis, which takes social concepts as exhaustive of social reality, "blinds us to the possibility of, for instance, social control through the employment of absurd, ambiguous, inconsistent or unintelligible doctrines ... we can sometimes only make sense of the society in question by seeing how the manipulation of concepts and the violation of categorial boundaries help it to work.'" This observation is also the basis of ]iirgen Habermas's critique of hermeneutics. For Habermas the hermeneutic grasp of the self-

understandings embedded in social praxis is an absolutely necessary moment of social analysis, but it is not sufficient. For it is also necessary to produce general knowledge about institutional relations-social power-and in so doing to remain sensitive to "when they express ideologically frozen relations of dependence that can in principle be transformed."'o Social agents necessarily possess some concepts of their power relationships. But they do not therefore necessarily possess all of the concepts sufficient to fully understand the determinants of their activity. An understanding of social power, then, must extend beyond a concern with shared concepts to a concern with structural relations. Such a mode of analysis is not only necessary to generate adequate knowledge; it also contains, as Habermas emphasizes, a critical promise, serving to potentially demystify social reality and releasing agents from "hypostatized powers."

Social Power and Social Structure Social power must be understood relationally. By relationa!!y ! do not mean, as in empiricism, in terms of contingent behavioral regularities, but rather in terms of the real underlying social relations that structure behavioral interaction. The relation between teacher and student, for example, is not a contingent relation between two parties who happen to encounter one another. It IS a historically enduring relation, the nature of which is precisely that teachers have students and vice versa; it is the nature of these social identities to be in relation to one another. And, as such, it is their nature to possess certain powers, powers that simply cannot be conceived as contingent regularities between the behaviors of teacher and student. The teacher possesses the power to design the syllabus, to speak in front of the class and direct classroom activities, and to give and grade assignments; the student possesses the power to attend class, to do the schoolwork, and to earn academic credit for his or her work. The capacities to perform these activities lOJiirgen Habermas, Knowledge and Human Interests (Boston: Beacon, 1971),

9Ernest Gellner, "Concepts and Society," in Wilson, Rationality, pp. 43-47. See also Alasdair MacIntyre, «Is Understanding Religion Compatible with Believing?" in Wilson, Rationality.

p. 310. See also Anthony Giddens, "Habermas's Critique of Hermeneutics," in his Studies in Social and Political Theory (New York: Basic Books, 1977), pp. 135-

64.

80

Empiricism, Realism, and Power

are part of the nature of the relationship; and their performance involves the drawing upon and the exercise of powers. The possession of these powers is a necessary feature of the structure of education; but the successful exercise of these powers is contingent. To be a teacher is to possess the power to direct the classroom, but teacher A may not succeed in directing her classroom, and the class may be unruly. Her power as a teacher, however, is not nullified by this particular failure. A particular teacher's consistent failure to direct the classroom is a different story, and it may well nullify her power. We would then likely say that she was a bad teacher, unsuited to the role of teacher and personally unable to exercise the social powers required by the role. And, more generally, the persistent inability of teachers in general to successfully direct their classrooms may very well indicate that the teacher-student relationship is in crisis and that students are exercising their powers to contest the structure of the relationship. I will thus define social power as those capacities to act possessed by social agents in virtue of the enduring relations in which they \ participate. Giddens distinguishes between a broad sense of power, the capability of an actor to intervene, and a narrower sense, "'the

capability to secure outcomes where the realization of these O\l!comes depends On the agency of others.,,11 What 1 have defined as social power refers to the more restricted sense. Thus, while the word power is properly used to describe many circumstances-for instance, my neighbor's persuasive ability, which resides in his .45 Magnum-the phrase "social power" is intended to call attention to the way capacities to act are distributed by generalized and enduring social relationships (is my persuasive neighbor also a policeman?). 1 am not proposing to univocally define the meaning of power, nor to stipulate a single acceptable use of the concept in social analysis. 12 What I am suggesting is that a generalizing social science takes as its primary object of study those enduring social relationships that distribute power to social agents. Individuals certainly llGiddens, Central Problems, p. 93. uFor a discussion of the different uses, and "grammars" of "power," see Peter Morriss, "The Essentially Uncontestable Concepts of Power," in Michael Freeman and David Robertson, eds., The Frontiers of Political Theory: Essays in a Revi~ talized Discipline (New York: St. Martin's, 1980), pp. 198-232.

The Concept of Power Revisited

81

possess idiosyncratic powers. But what makes these socially significant is the way they are implicated in more enduring relation- . ships. It makes perfect sense to claim that "David Rockefeller is a powerful man." But a social theory of power must explain what kinds of social relations exist and how power is distributed by these relations, such that it is possible for David Rockefeller to have the power that he has. To do this is not to deny that it is he who possesses this power, nor to deny those personal attributes determining the particular manner in which he exercises it. It is simply to insist that the power individuals possess has social conditions of existence, and that it is these conditions that should be the primary focus of theoretical analysis. To propose such a structural understanding of social power is not to detach the concept of power from human agency. As we have seen in Chapter 2, the understanding of social structure I rely on rejects any bifurcation of structure and agency. As we have defined it, social power refers to the capacities to act possessed by agents in virtue of their social relations. And what are these relations but idioms of human conduct? To say that teachers and students are in a structural relationship is only to say that there are people called teachers and students who characteristically do the things the relationship involves, who characteristically exercise those powers distributed by their roles. A social power that is never exercised can hardly be said to exist. But its exercise is always shaped and constrained by pregiven relations. I am going to school this afternoon to give a lecture on Machiavelli; and in doing so, however unintentionally, I am exercising the power of a teacher. It is this power that enables me to give a lecture to a class full of students rather than to an empty room (imagine how quickly campus security guards would cart away someone off the street who, claiming to know about Machiavelli, walked into a classroom full of students waiting to hear their teacher's lecture), and also constrains me from doing otherwise (I would really much prefer to sit in the fields beneath an oak tree, reflecting upon virtue with a group of dedicated students). The sorts of structurally distributed powers of which we speak are chronically exercised in the course of everyday life, at home, at work, at school, at the tax collector's office. And the successful exercise of them is always contingent. Bosses by nature have the

82

Empiricism, Realism, and Power

power to supervise production, but tomorrow the workers may strike. Teachers by nature have the power to conduct class lessons, but tomorrow the students may boycott class and conduct their own teach-in. It is a real, necessary feature of the currently existing structure of education that teachers possess certain supervisory powers. But the exercise of these powers, the way the teacherstudent relationship is worked out in practice, is contingent, determined by the way particular individuals and groups choose to deal with their circumstances. Thus, as Ted Benton has characterized the realist view: "The advocated position is systematically determinist and causal, but at the same time it does not reduce agency to a mere 'bearer' of the activity of extrinsic structures. H13

The relational concept of power we have outlined above is defended by Terence Ball in an important essay. 14 Ball argues that the empiricist view of power as a contingent behavioral relation is impossible if certain social scientific concepts are to be taken seriously. This is because the meanings of these concepts entail certain relations of power. Ball, drawing heavily upon Hegel, insists that certain terms are necessarily relational. The terms of the relation between master and slave, which is the paradigm of all power relations, are a case in point. Ball writes: "Since (as Hegel certainly recognized) A would not even be A without B, then the relationship between them is conceptual and not contingent ... is not part of the meaning of being a 'master' to have a 'slave' to do your bidding? ... to be a master means to be able (or in a position) to command and be obeyed. That, after all, was the point of Hegel's excursis on lordship and bondage: the master's very identity as master is (logically) dependent upon the slave's continued subservience."15 Because the master-slave relationship is not contingent but conceptually necessary, Ball concludes, on the empiricist view it is not a relation of power. The empiricist concept of power is thus, as Ball writes, "reduced to absurdity."" Ball, in line with post-positivist philosophy, makes a convincing 13Ted Benton, "Objective Interests and the Sociology of Power," Sociology 15

(May 1981), 179. 14Terence Ball, "Two Concepts of Coercion," Theory and Society 15 (January 1978),97-112. "Ibid., pp. 106-109. 16Ibid., p. 108.

The Concept of Power Revisited

83

case for the theory-ladenness and mutual implication of theoretical terms. The concept of master entails the concept of slave, and the relation between these concepts is thus necessary. It is very important to insist that our theoretical language for talking about social power fails to conform to Humean canons. We must, however, pnsh this argument a bit further. Power is not just conceptually necessary in the above sense; it is really necessary. The concept of master entails power over slave. But the real master, to whom this concept refers, does not have power because the concept of master means that he does; he has power because he is an element of a real, structural relationship with real slaves. The conceptual necessity Ball exposes is grounded in the real social necessity of the master-slave relationship. Ball in fact comes close to asserting this when he writes that the relationship between master A and slave B, being conceptually necessary, is not for empiricism a relation of power because it fails to "qualify as causal in the contingent Humean sense."" Ball is satisfied to perform a reductio ad absurdum, demoustrating the incompatibility of empiricist notions with commonly accepted social scientific terms. But he also suggests the possibility that the master-slave relationship, while not causal in the Humean sense, is causal in another, realist sense.

Power and Domination The concept of domination does not figure in the three faces of power controversy. Instead, all power is assimilated to "power over," which is construed as a behavioral regularity. This view of "power over" suffers from two basic deficiencies that the realist view avoids. It deprives the notion of "power over"-domination-

of its specific significance, and it also provides a misleading picture of domination as a causal relation. In the realist view power, as those capacities implicated in ongoing social practices, is a necessary feature of social interaction; where there is society there is social power. However, "power over," or domination, is not therefore a necessary feature of society. Following Giddens, I define domination as the asymmetrical dis17Ibid.

84

Empiricism, Realism, and Power

tribution of social power." Relations of domination and subordination comprise a subset of power relations, where the capacities to act are not distributed symmetrically to all parties to the relationship. The term domination comes from the Latin dominus, or lord; and it has obvious connections with the mastery of the household, as it does with the desmegne/dominion/domain of the feudal lord (the concept of domain here designating the necessary connection between a social space and a social prerogative). It means, variously, mastery, control, command, or prevalence.!' The general sense of its meaning is clear. By asymmetry we do not simply mean social interdependence or a division of labor. Medieval butchers and bakers, for instance, perform different occupations and are mutually dependent on one another; but they are symmetrically related by mutual need. They possess equivalent powers of selfemployment. And factory workers on an assembly line are similarly dependent on one another, but they too are symmetrically related by mutual need and the shared possession of the capacity to work. These are relations of difference, but not necessarily of domination. But the Roman lord does not simply perform a different set of activities from those of his wife and his slaves; he directs the house~

hold, thereby directing and controlling the conditions under which his wife and slaves perform their activities. Similarly, as we shall see, the capitalist directs the labor process, controlling the conditions under which wage laborers labor. The concept of domination thus refers neither to a contingent regularity nor to mere social difference; it refers to a structurally asymmetrical relationship, whereby one element of the relationship has power over another in virtue of its structural power to direct the practices of the other. This is what Max Weber seems to have meant when he defined domination as the "authoritarian power of command." He writes "To be more specific, domination will thus mean the situation in which the manifested will (command) of the ruler or rulers is meant to influence the conduct of one or more others (the ruled) and actually does influence it in such a way 18Giddens, Central Problems, p. 93; Anthony Giddens, A Contemporary Cri~ tique of Historical Materialism (Berkeley: University of California Press, 1981j, pp.50-52. 19See "Domination," Oxford English Dictionary, vol. 2 (Oxford: Clarendon, 1933).

The Concept of Power Revisited

85

that their conduct to a socially relevant degree occurs as if the ruled had made the content of the command the maxim of their conduct for its very own sake. Looked upon from the other end, this situation will be called obedience."'o Weber's conceptualization of domination was an early causality of behavioralism in political science, having been co-opted into the attempt to "operationalize" the concept of power. 2 ! This is understandable insofar as Weber, operating within the tradition of methodological individualism, emphasizes the dimension of will and intentionality Of, in Lukes's terms, action. Moreover, when Weber writes of "the causal chain extending from the command to the actual fact of compliance,"" he offers what appears to be an unmistakably Humean formulation. But his definition also suggests a structural dimension, conceiving of domination as a situation or condition rather than an event. He makes this clear by treating modern bureaucracy as the paradigm of domination, and by providing as further examples the village chief, the banker, the judge, the craftsman-socially positioned roles that, by their nature, possess certain powers over subordinates. The language he usescommand and obedience, ruler and ruled-is equally suggestive of a structural approach. A realist view is not simply able to appreciate the structural nature of "power over" which is occluded by empiricism; it also provides a more plausible account of what kind of causal relation is involved in relations of domination. On the empiricist view to say that master has power over slave, or dominates slave, means that the behavior of the master causes the behavior of the slave, that is, it regularly antecedes it. According to the realist position it is not the behavior of the slave that is caused by the behavior of the master; rather, the master-slave relationship is the material 2°Max Weber, Economy and Society, vol. 2, ed. Guenther Roth and Claus Wittich (Berkeley: University of California Press, 1978), p. 946. 21See, e.g., Robert A. Dahl, "Power," International Encyclopedia of the Social Sciences, vol. 12 (New York: Macmillan, 1968). We should not overlook Talcott Parsons's notorious translation of Weber's herrschaft as "imperative coordination" rather than domination. For a good discussion of this issue, see Randan Collins, "A Comparative Approach to Political Sociology," in Reinhard Bendix et al., eds., State and Society: A Reader in Comparative Political Sociology (Berkeley: University of California Press, 1973). 22Weber, Economy, p. 946.

86

Empiricism, Realism, and Power

cause of the behavior of both the master and the slave, and the specific way the master and the slave (an equally purposive agent) choose to act out the relationship is the efficient cause. For empiricism there are only efficient causes; but the consequence of this IS an mablhry to explain both the mutual identities and the capabdltles avatlable to parties to social relationships. On my view the master, whde dommant, operates under constraints just as the slave does. And the slave, though subordinate, is also causally effective m determmmg the outcomes of his encounters with his master. Nlcos Poulantzas makes a similar point in discussing the class reiatlOns of capitalism: "The field of power is strictly relational. ... The place of each class, and hence its power, is delimited (i.e., at once designated and limited) by the place of other classes. Power IS not attached to a 'class in itself' understood as a collection of agents, but depends upon, and spring from, a relational system of matenal places occupied by particular agents.,,2; Poulantzas, and the Althusserian tradition more generally, emphasizes the determmacy of the "relational system of material places" at the expense (though not to the exclusion) of the effectiviry of the agents who e:-:erClse their structurally given powers (See Part II). What is pertment here, though, is his insistence on the structural constitution of all power, including the power of dominant groups. . There are a number of important consequences of the realist view. Whereas the empiricist concept of power metaphorically represents dommanon as a lateral relation between the behaviors of social agents, realism views domination as a vertical relation between the agents themselves. This view takes much more seriously the rea!tnes of command and obedience as enduring features of social life differing in kind from the more contingent encounters of everyday life. It also refuses to conllate power and domination. While power is a necessary feature of social existence domination is not. And while it may very well be true that huma~ history is a history of structures of domination, it does not follow that this is the way it has to b~. To think of power as a causal concept is not necessanly to Imagme that there will always be some people possessmg power over others. Finally, on the realist view the practical outcomes of relations 23Nicos Poulantzas, State, Power, Socialism (London: Verso, 1980), p. 147.

The Concept of Power Revisited

87

of domination are open-ended, determined both by the structural powers accruing to social agents and the particular ways they exercise these powers. It is to this issue that we now turn.

Reciprocity and the Negotiation of Power While on the empiricist view power is a one-way relationship of behavioral causation, on the realist view we have developed relationships of power necessarily involve reciprociry. Giddens writes: "However wide the asymmetrical distribution of resources mvalved, all power relations manifest autonomy and dependence 'in both directions.' ,,24 Thus even the power of a slaveowner IS Circumscribed by the structure of the master-slave relationship, more specifically its basis in slave labor and all of the normative and cultural elements surrounding this. The power the slaveowner has to buy and sell slaves, to control the conditions of their activiry, and to direct and exploit their labor is articulated with the minimal powers slaves possess over their bodies so that they can perform those services defined by their relation to the master. Roderick Martin, in The Sociology of Power, takes antebellum American slavery to be an extreme example of the one-sided character of power, based upon the "total power" of the master. In support he produces an interesting text, a Virginia slave code of 1669 legalizing the killing of slaves by masters: "It cannot be presumed that the propensed malice (which alone makes murder felony) should induce any man to destroy his own estate. "25 Martin, however, seems to miss the point of this code, which can only be ascertained by inquiring into the relation between the laws and the practices of slavery. The legal right to violently punish slaves seems here quite explicitly articulated with the real functIon slaves performed. As real property, as exploited labor, there was no reason to assume that a master would simply murder a slave for no good reason (keep in mind, of course, that the master's reasons need not be our reasons; as C. Wright Mills pointed out long ago, SOCial 24Giddens, Central Problems, p. 149. 25Roderick Marrin, The Sociology of Power (London: Routledge & Kegan Paul, 1977), pp. 59-63.

88

Empiricism, Realism, and Power

rationalitie~ are not necessarily truly reasonable). The rationality of slavery mvolved the requirement of a reproduced slave labor force. Of course, the fact that the master possessed legally unlimited powers of vIolence IS crucial and serves to highlight that slavery is a relatiOnShIp based on extreme asymmetries of power. But even here the actual role of violence, as a social practice, was determined by the structure of the master-slave relationship and the distribution of powers to masters and slaves. This is not to deny the brutality of slavery, and it is certainly not to offer a suhstantive historical analysis of its modes of operatIOn: In fact, It IS quite possihle that some masters did wantonly kIll theIr slaves. The point is simply that, in doing so, they would have accomplIshed the same thing that capitalists accomplish when they squander their capital-undermined the very basis of their pOSitIOn of domination. What Marx writes about capitalism could Just have easily been said about slavery by substituting slave for capItal and worker and master for capitalist: "Capital in its beingfor-Itself IS the capItalIst.... I may well separate capital from a gIven mdividual capitalist, and it can be transferred to another. But m losmg capital, he loses the quality of being a capitalist. The capItal IS mdeed separable from the individual capitalist but not from the capitalist, who as such confronts the worker. "2: To be a master is to have certain powers in relation to slaves and vice versa: Just as the imperatives of capitalism place limits o~ what a~y typIcal capItalist can do with his or her capital, the ImperatIves of slavery, and its basis in slave labor, place limits upon what any rational master can do. It is possible of course for ~a~ti~~lar mas,~ers to o:erst~p t~ese bounds, to ab~se a power ~hat, m Its normal operatIOns, IS VIOlent and dehumanizing enough.27 But 10 do1Og so, they would be, so to speak, consuming their power through its exercise; a power that, under the conditions of slave society, while separable from this particular master, is inseparable from the master as such. The structural limits of power are important both because they define the speCIfic nature of power relations (e.g., master-slave ver26Karl Marx, Grundrisse, trans. Martin Nicolaus (New York: Vintage, 1973), p.303.

2~On the dehuman~zing character of slavery, see Orlando Patterson, Slavery and Soczal Death (Cambndge, Mass.: Harvard University Press, 1979).

The Concept of Power Revisited

89

sus official-citizen versus capitalist-worker), and because they shape the contours of the life chances of agents, of social conflict, and of the possibilities for social change. To emphasize that power relations always involve some element of reciprocity is not to deny the salience of domination. It is to recognize that domination is always of a specific sort, and, moreover, that it involves the mutual performance of activities. This recognition also enables us to incorporate directly into the conceptualization of power the chronic negotiation attending its exercise. Giddens calls this the" dialectic of control." He writes: "However wide-ranging the control which actors may have over others, the weak nevertheless always have some capabilities of turning back resources against the strong.,,2' This emphasis on the mutuality of power relations, and on the genuine agency of the subordinate, is an important theme within the phenomenological tradition. Thus in Hegel's famous chapter "Lordship and Bondage" in The Phenomenology of Mind, while the slave appears to the master as "inessential," a wholly dependent instrument of the master's will, it becomes "evident that this object does not correspond to its notion"-that it is the master who is dependent on the slave's activity. (Of course, in Hegel this insight is incorporated within a philosophy of history, and thus for him "just as lordship showed its essential nature to be the reverse of what it wants to be, so, too, bondage will, when completed, pass into the opposite of what it immediately is: being a consciousness repressed within itself, it will enter into itself, and change atound into a real and true independence. ,,)29 Georg Simmel also insists on the effectivity of the subordinate. He writes: Nobody, in general, wishes that his influence completely determine the other individual. He rather wants this influence, this determination of the other, to act back upon him. Even the abstract willto-dominate, therefore, is a case of interaction. This will draws its 28Giddens, Contemporary Critique, p. 63. 29G.W.F. Hegel, The Phenomenology of Mind, trans. J. B. Baillie (New York: Harper Torchbooks, 1967), pp. 229-40. For discussion, see Alexandre Kojeve, Introduction to the Reading of Hegel, trans. Allan Bloom (New York: Basic Books 1969), chap. 1; and Jean Hyppoiite, Genesis and Structure of Hegel's Phenomen· oiogy of Spirit (Evanston, Ill.: Northwestern University Press, 1974). For a critical analysis of Hegel's idealism in this section, see Chris Arthur, "Hegel's Master! Slave Dialectic and a Myth of Marxology," New Left Review, no. 142 (NovemberDecember 1983), 67-75.

90

Empiricism, Realism, and Power

satisfaction from the fact that the acting or suffering of the other, his positive Or negative condition, offers itself to the dominator as

the product of his wilL ... The practical function of this desire for domination ... is not so much the exploitation of the other as the mere consciousness of this possibility .... But still, even the desire for

domination has SOme interest in the other person, who constitutes a value for it .... Even in the most oppressive and cruel cases of subordination, there is still a considerable measure of personal freedom.-'o

These formulations are characteristically idealist, treating power as an outgrowth of pure consciousness or will; but they insist on something ignored by empiricism-that power operates dialectically and not accordmg to a Newtonian model of stimulus and response. Giddens, in his discussion of this issue, emphasizes the dimension of free will in social life, the fact that, barring the confinements of a literal or figurative straitjacket, the abiliry to say no is always a live option. "This," he writes, "accounts for the intimate tie between agency and suicide. Self-destruction is a (virtually) always open-option, the ulnmate refusal that finally and absolutely cancels the oppressive power of others; hence suicidal acts themselves ca.n be understood as concemed with the exercise of power.,,31 This is an Important point insofar as it forces us to see all social outcomes as the contingent results of purposive human activiry; and while it emphasizes the importance of choice (and, correlatively, of moral responsibiliry), it also recognizes the determinacy, as well as the possible oppressiveness, of circumstances and conditions not of our own choosing. But it also seems to be an incomplete account of the way the dialectic of power-the capabiliry of the weak turning back resources agamst the strong--operates in social life outside of extreme situations like muggings and death camps.32 Giddens J°Georg Simmel, The Sociology of Georg Simme~ ed. Kurt Wolff (New York:

Free Press, 1950), pp. 181-82.

J1Giddens, Central Problems, p. 149. HHannah.Arendt, for instance, clarifying her views on the moral responsibility of those Jewish le~d~~s who collaborated with Nazi policies, wrote: "I said that ther~ w~~ no pOSSlblilty of resistance, but there existed the possibility of doing nOt~l.ng. She was, however, very careful to distinguish the "limited freedom of ?eClSlOn and action" available in the ghetto with the "immediate pressure and lmp~ct of.terror" within the concentration camp ("Letter to Gershom Scholem," repnnted 10 Hannah Arendt, The Jew as Pariah, ed. Ron H. Feldman [New York: Grove, 1978J, pp. 248-49).

The Concept of Power Revisited

91

himself later observes: "Most circumstances of control, of course, are not nearly so all-embracing as those of captor and captive.,,·l3 In this sense Hegel's dialectic of master and slave is suggestive. The real dialectic of power in social life lies in the real social powers possessed by subordinate parties to enduring relationships, and in the reciprociry of social practice. The capacities possessed by the slave, which can be turned against the master, do not reside simply in the freedom of human agency as such; they reside in the defimte social powers distributed by the master/slave relationship-plowing the fields, picking the cotton, and the like. The master-slave relationship, like all social relationships, is chronically negotIated and renegotiated on the basis of this reciprocal possession of powers. The "normal" practices of slavery distribute certain supervisory, disciplinary, and appropriative powers to masters and laboring powers to slaves. In the "normal" working out of this relationship masters tend to perform directive activities, and slaves laboring activities. But though this distribution of powers is a necessary feature of the structure of slavery, the successful exercise of them is a contingent outcome of the interaction between concrete masters and slaves. These outcomes depend on the respective abilities of the parties to mobilize the resources at their disposai and to use them effectively. The fact that social power characteristically involves relations of mutual dependence allows for a variery of modes of leverage, maneuvering, and strategic bargaining between agents. Of course the forms the negotiation takes will vary according to the structure of the relationship. Industrial workers in advanced capitalist societies, for instance, have recourse to legal rights and a judicial machinery in negotiating their power with capitalists. Antebellum American slaves had no such recourse, and thus for them resistance and negotiation characteristically took more covert forms.34 Giddens points out that Erving Goffman's distinction between "front" and "back regions" illuminates one frequent mode 33Giddens, Contemporary Critique, p. 63. J4See John W. Blasingame, The Slave Community: Plantation Life in the Ante~ Bellum South (Oxford: Oxford University Press, 1972); Kenneth Stampp, The Peculiar Institution (New York: Vintage, 1956), esp, pp. 322-82; and Herbert W, Gutman, The Black Family in Slavery and Freedom, 1750-1925 (New York: Pantheon, 1976).

92

Empiricism, Realism, and Power

by which subordinates negotiate power: "Performances in front regions typically involve efforts to create and sustain the appearance of conformIty to normative standards to which the actors in question may be indifferent, or even positively hostile when meeting in the back. ,,35 Goffman's writings emphasize the v~rious games whereby everyday life-even in "total institutions" like asylums and pnsons, wher~ domi~ation takes an extreme form-is negotiated by Its partIcIpants. John Blasingame's The Slave Community: Plantation Life in the Antebellum South similarly examines the dIfferent responses and forms of resistance to slavery evidenced by slaves. He concludes: "Rather than identifying with and submlttmg totally to his master, the slave held onto many remnants of hIs Afncan culture, gained a sense of worth in the quarters, spent most of his tIme free from surveillance by whites, controlled important aspects of his life, and did some personally meaningful thmgs on hIs own volition. This relative freedom of thought and actIOn helped the slave to preserve his personal autonomy and to create a culture which has contributed much to American life and thought. ,,37 Much of the "new social history" is concerned with the subcultures and modes of self-organization of subordinate groups-peasants, workers, women, blacks and other racial groups-and the way in which these embody and facilitate resistance to forms of domination. 38 What bears emphasis is that these forms of negotiation are not as behavioralist formulations might suggest, situations from which power springs, as it were, ex nihilo. They are situations in which the exercise of structurally pregiven powers is negotiated and contested. It is at this level that the question of compliance-getting people to do thmgs-can be broached. On the view I have developed power relations are not compliance relations. They are endurmg relatIOns of reproduced, reciprocal practices. The exercise 35Giddens, Central Problems, p. 208. 36See Erving GoHman, The Presentation of Self in Everyday Life (New York: ~ouble?ay, 1959); Asylums (New York: Doubleday/Anchor, 1961); and Relations m Publtc (New York: Harper Torchbooks, 1971). 37Blasingame, p. viii. J8See Barry Goldberg, "A New Look at Labor History," Social Policy (Winter 1982),54-63; and Gregor McLennan, Marxism and the Methodologies of History (London, Verso, 1981), pp. 112-28.

The Concept of Power Revisited

93

of power does sometimes involve problems of compliance, though often in social life powers are exercised rather matter-of-factly and unproblematically. Capitalists make investment decisions, workers work. Teachers give out assignments, students do them. Drill sergeants bark out orders, soldiers obey them. As William James once remarked, "habit is the enormous flywheel of society," and power is often (though, as we have emphasized, not always) exercised, and reproduced, rather routinely. Although power relations are not best conceived as compliance relations, problems of compliance are chronic features of social life. The reproduction of practices, and of powers, is always problematic. In this sense, the specific individuals and groups who draw upon structural powers and exercise them in accordance with their own concrete purposes-meeting this production deadline, getting my students to do that assignment-are always faced with the problem of achieving these purposes. The mobilization of resources, the exchange of threats and counterthreats, the offering of positive and negative sanctions, the calculation of costs and benefits-all these figure importantly at this level of strategic interaction. Exchange theories of power shed light on the inherently problematic and negotiated exercise of power, though the assumptions about rationaliry and the transparency of interests entertained by most exchange theorists are unwarranted and highly suspect (see the following section).39 It is this indeterminacy in the exercise of power that accounts for the openness of history we discussed in Chapter 2. While power is a pregiven property of structurally located agents, these structures are only relatively enduring, and the activities of subordinate groups can play a crucial role in altering them. Students, for instance, can strategically withdraw their learning powers, forcing school administrators to allow them a greater role in school governance; and workers can collectively dispose of their labor power by organizing a trade union. These are examples of ways in which social struggle can renegotiate the terms of social power. And they 39See Brian Barry, ed., Power and Political Theory: Some European Perspectives (London: Wiley, 1976), for some good discussions of exchange theory. For a more general criticism of rational choice theory, see Roy Bhaskar, The Po~sibility of Naturalism: A Philosophical Critique of the Contemporary Human Sctences (At~ lantic Highlands, N.J.: Humanities Press, 1979), pp. 137-46.

94

Empiricism, Realism, and Power

are also examples of the role solidarity and organization can play in enhancing the ability of particular agents to negotiate their interests. At the limit, however, are those instances in which relations of power are more fundamentally challenged and transformed. It is the nature of capitalism, as we shall see in Part II, that workers possess labor power that is put at the disposal of capitalists; but this is not to say that the practice of work could not be organized in different ways. It is the vision of socialist theory that workers, defined under capitalism as a subordinate class, might transform the structure of class relations, claiming for themselves powers previously in the hands of capitalists. In insisting that social power is a structural capacity we must not ignore the persistence, passion, and ingenuity with which living and breathing human beings exercise it, cope with it, and struggle to transform it. Once again, our intention is not to rigidly bifurcate social structure and the manifold forms of human agency, but to see how power is implicated in both, and in fact links them together. The analysis of power must examine those structural relationships that distribute the capacity to act. But it must also examine the concrete history whereby these relationships are maintained and changed, and the forms of organization of those groups whose activity makes these things happen. A structural approach to social power thus does not license an invidious essentialism, whereby particular struggles and forms of organization are seen as no more than epiphenomenal "expressions" of underlying relations. Barry Hindess, in a 1982 article, insists upon the "conditionality of outcomes" and argues that any concept of power as capacity must fail because "outcomes are not 'predictable' and 'unvarying' in the way this conception requires.,,40 The unpredictability of social outcomes is something about which we agree; there can be no "guarantees" that power will be successfully exercised on any particular occasion. And more generally, as Hindess points out, the negotiation of power involves "struggles, not the playing-out of some pre-ordained script."41 But Hindess equivocates between suggesting that (1) the exercise of power is contingent (he writes that 4°Barry Hindess, "Power, Interests, and the Outcomes of Struggles," Sociology 16 (November 1982), SOl. 4JIbid., p. 506.

The Concept of Power Revisited

95

outcomes are "rarely the simple products of initial conditions" [italics mine l; also note the scare quotes when he talks about the "unpredictabiliry" of power), and (2) initial (structural) conditions are entirely irrelevant, there being nothing more than the exerCIse of power. The first suggestion is sound, and Hindess does well to remind us of the importance of struggle. The second is groundless, not only unhinging struggle from its conditions of existence, but rendering unintelligible what parties to struggle are strugglmg about. Michel Crozier, in The Bureaucratic Phenomenon, has some very interesting things to say about the uncertainties surrounding the exercise of power in highly rationalized organizations, which compel even the most powerful officials to "bargain and compromise with all the people whose cooperation is indispensable at each level.,,42 Yet, Crozier insists, "One should not translate the logic of the struggle into an overly black-and-white picture.... Other forces are operating which insure the minimum of consensus and organizational commitment that prevent people from extracting too much from, or being too much exploited in, their reciprocal de:~ls.. "43 Crozier's comments here caution us about heing so sen~i~;~~' to the role of struggle that we forget those structural conditions under which struggle takes place. They also raise another important question, to which we now turn-the question of interests, their constitution, and the problem of consensus.

Power, Interests, and Ideology Lukes, we may recall, introduced the concept of interest into the three faces of power controversy. For Lukes, the concept of interest is necessary to the discussion of power insofar as it answers the qnestion of the counterfactual-what would B do were it not for A's behavior? We have seen that this way of thinking about power is mistaken; and Lukes's counterfactual question does not figure in my realist account of power insofar as I reject the Newtonian 42Michel Crozier, The Bureaucratic Phenomenon (Chicago: University of Chicago Press, 1964), p. 163. 4}Ibid., p. 167.

96

Empiricism, Realism, and Power

premise on which it rests. Rather than A getting B to do something B would not otherwise do, social relations of power typically involve both A and B doing what they ordinarily do. The structure of education, not teachers, causes students to act like students and teachers to act like teachers. Teachers and students, as the social identities that they are, would not "otherwise do" anything but what teachers and students tend to do. And neither a conflict of revealed preferences, nor of objective interests, must be discovered in order to attribute power to these roles. As Lukes recognizes, a relation of power can exist even in the absence of a conflict of revealed preferences. Contrary to Lukes, however, a relation of power can also exist in the absence of a conflict of "objective interests." It might very well be the case that my power over my students is in their best interest; but the relationship is not for that reason any less one of domination and subordination. Lukes's own formulation would seem to deny this, defining away any form of power that seems or is claimed to be in the mutual interest of the parties involved, and thus opening him up to the charge of "vanguardism. ,,44 But to claim that the concept of power does not require recourse to the concept of interest in the way that Lukes argues is not to deny that the concept of interest plays a necessary role in the analysis of power. In acknowledging its centraliry, however, we must also clarify the precise meaning of the concept. I would suggest that there are at least three different meanings of the concept of interest that must be clearly distinguished. The first meaning of "interest" refers to the subjective interests, or revealed preferences, that are actually held by particular agents. We have suggested that in this sense the concept is not epistemically necessary to theoretical claims about social power. Different individuals have different preferences. Some may like their social role. Some may not. Some may not and yet prefer to do nothing about it, possibly because it is to them a means to other more preferential 44See Peter Abell, "The Many Faces of Power: Revealed Preference, Autonomy, and Teleological Explanation," Sociology 11 (January 1977), 3-23; K. Thomas, "Power and Autonomy: Further Comments on the Many Faces of Power," Sociology 12 (May 1978), 332-35; and G. W. Smith, "Must Radicals Be Marxists? Lukes on Power, ContestabiHty, and Alienation," British Journal of Political Science 11 (1981),405-25.

The Concept of Power Revisited

97

ends, like making money or graduating school. We can talk theoretically about the structure of power in the classroom without reference to the preferences of the students, who are subordinate even if they prefer to remain so. But while social relationships are not simply effects of subjective interests, being themselves important determinants of preference formation, subjective interests are crucially relevant to the exercise and the reproduction of social power. As we have insisted, the exerCise of power is contingent, and the outcomes of its exercise are determined both by the structural distribution of power and the subJective understandings, preferences, and "will" of concrete SOCial agents. Theoretical claims about the structural location of power may not require, as evidence, reference to the subjective interests of particular agents. But the analysis of the working out of relations of power, of their concrete reproduction and transformation,. necessarily involves attention to the preferences and strategic obJectives of socially situated individuals, groups, and organizations. A realist view of power does not reify social agents and their specific preferences. But neither does it take them as a given. Rather, it suggests that they be expiained with reference, though not reduced, to the relatively enduring relationships in which agents participate. The second meaning of "interest" is that suggested by Lukesobjective interest, or what is really in the interest, or good, of an 45 agent, whether he or she thinks so or not. I argued in Chapter 1 that we need not have recourse to this concept in order to Justifiably identify a relation of power, and thus the peasant with a gun t? her head is subordinate to the commissar of agriculture even If collectivization is in her interest. While it is thus possible to talk about power independently of the issue of objective interests, it does not follow that the concept is unintelligible or irrelevant, as many of Lukes's critics have claimed. We will return to this issue shortly. These two usages of "interest"-subjective and objective-have preoccupied theorists in the debate on the concept of power. I Will introduce a third meaning of interest--constltutlve, or real tnterest. 45See William E. Connolly, "On 'Interests' in Politics," Politics and Society 6Ralph Miliband, "Poulantzas and the Capitalist State," New Left Review, no, 82 (November-December 1973), 87.

162

Marxism and Power

state is a site, a "strategic field," of the exercise of power, and that political struggles are "inscribed in its framework." The denial that the state has power as such should be read as an insistence that the effects of state power are never certain, however much the state tends to reproduce class relations. Jessop clarifies this when he claims that "the state is a set of political institutions that cannot, qua institutional ensemble, exercise power. "37 The powers of the state are always exercised separately, by particular officials, occupying specific institutional roles. Moreover, they are exercised conjuncturally. Political struggles upon, within, and between state institutions will always determine how state power is exercised in, for example, a specific law, policy, or administrative action. As Miliband has pointed out, it would be much clearer to simply subscribe to a distinction between class power and state power, viewing the latter as "the main and ultimate--but not the onlymeans whereby class power is assuted and maintained."" On this view (which, if I am correct, is implicit in the awkward formulations of Therborn and Poulantzas) state power is understood as the various public powers of territorial rule and physical coercion pos-

sessed by the specialized agencies of centralized government. Class powers, as we have seen in the previous chapter, are understood as those powers, structural and organizational, which are possessed by economic classes. As Miliband points out, the relationship between state power and class power constitutes the central problem of Marxian political theory." While the writings of contemporary Marxists are only intelligible if we preserve the distinction between state power and class power, they are also only comprehensible if we recognize the centrality of the Marxian interest in the causal relationships that obtain between them. Poulantzas's remark on state power is an insistence upon the futility of any view of the state that fails to analyze its implication in the class relations, and class struggles, with which it is associated.

37Jessop, Capitalist State, p. 221. 38Miliband, "Poulantzas," pp. 87-88. See also Miliband's discussion in his Marxism and Politics (Oxford: Oxford University Press, 1977), pp. 54-55, 66-

67. J9Miliband, Marxism, p. 68.

Marxism and the State

163

The Relative Autonomy of the Capitalist State The concept of relative autonomy has played an important role in recent Marxian political theory, designating the causal relations between class domination and the state. Poulantzas, for instance, drawing upon Marx's analysis of the separation of state and civil society under capitalism, writes of the "characteristic autonomy of the economic and the political" which distinguishes capitalism from feudalism. While the latter was based upon the localized political power of the lords, which was implicated directly in the relations of production, capitalism is based upon a sphere of institutionally separate economic relations that are secuted by a legal and coercive order outside of itself. Poulantzas writes of the capitalist state: Its fundamental distinctive feature seems to be the fact that it contains no determination of subjects (fixed in the state as "individuals," "citizens," "political persons,") as agents of production; and trus was not the case in other types of state. At the same time this class state presents a further specific feature: namely, that political class domination is constantly absent from its institutions. It presents itself as a popular-class state. Its institutions are organized around the

principles of the liberty and equality of "individuals" or "political will implied by the monarchical principle, but on the ensemble of formally

persons. H Its legitimacy is no longer founded on the divine

free and equal citizens and on the popular sovereignty and secular

responsibility of the state towards the people .... The modern capitalist state thus presents itself as embodying the general interest of the whole society, i.e., as substantiating the will of that "body politic" which is "the nation."40

The capitalist class, then, unlike the dominant classes in other modes of production, is not a ruling class. 41 The capitalist does 4°Poulantzas, Political Power, p. 123. 41While the recognition of this point goes back to Kautsky, C. Wright Mills perceptively remarked on it in The Power Elite (Oxford: Oxford University Press, 1956), p. 277. Mills writes: " 'Ruling class' is a badly loaded phrase. 'Class' is an economic term; 'rule' a political one. The phrase, 'ruling class,' thus contains the theory that an economic class rules politically. , .. Specifically, the phrase 'ruling class,' in its common political connotations, does not allow enough autonomy to the political order and its agents." Richard W. Miller, in his Analyzing Marx: Morality, Power, and History (Princeton, N.].: Princeton University Press, 1984),

164

Marxism and the State

Marxism and Power

rule, so to speak, the production process. But political rule, or territorial control, is ultimately vesred in the officials of the state, who act in the name of a generalized interest. The prerogatives of class domination are, as Poulantzas argues, absent from the institutions of the capitalist state. This means that a theory of such a state can in no way be reduced to an analysis of the economic or to any notion of the state as a reflection or expression of the economic. The state is a specific and relatively autonomous structure. As Poulantzas puts it, it is precisely this which "permits us to constitute the political as an autonomous and specific object of science. H42 Jessop has pointed out that the concept of relative autonomy has served to confuse as much as to clarify theoretical discussion, and he goes so far as to suggest that it should be "abandoned ... along with other Althusserian notions."" He quotes Poulantzas's response to the question of "how relative is the relative autonomy of the state": I cannot reply to this question, since in this form it is utterly absurd. I could have only answered this question, couched in these general terms, if I really had been guilty of structuralism. I can give no general answer-oot, as Miliband believes, because I take no account of concrete individuals or of the role of social classes, but precisely because the term "relative" in the expression "relative autonomy" of the state (relative in relation to what or whom?) here refers to the relationship between the state and dominant classes (i.e., relatively autonomous in relation to the dominant classes). In other words, it refers to the class struggle within each social formation and to its corresponding state forms. True, the very principles of Marxist theory attempt.;; to defend a "ruling class" interpretation of Marxism, but he does so without taking any account whatsoever of contemporary Marxian political theory. Miller in fact relies almost exclusively on the one text of Marx-The Eighteenth Brumaire of Louis Napoleon-which has been read by most commentators as providing the weakest foundations for such an interpretation; and his offhanded remark about the "increasingly sterile controversy between the so-called 'structuralist' and 'instrumentalist' interpretations" (p. 105) simply indicates, in my view, the limitations of his own argument. 42Poulamzas, Political Power, p. 29. 43Bob Jessop, Nicos Poulantzas: Marxist Theory and Political Strategy (New York: St. Martin's, 1985), p. 136. Jessop's account of the logic and genesis of Poulantzas's thought is extremely valuable, and it supports the realist reading I provide here.

165

lay down the general negative limits of this autonomy. The (capitalist) state, in the long run, can only correspond to the political interests of the dominant class or classes .... Yet, within these limits, the degree, the extent, the forms, etc. (how relative, and how is it relarive) of the relative autonomy of the state can only be examined (as I constantly underline throughout my book) with reference to a given capitalist state, and to the precise conjuncture of the corresponding class struggle. 44

Jessop rightly observes that the concept of relative autonomy functions here to designate both the complex mechanisms of class power and struggle and the ultimate necessiry of the reproduction of class domination. He criticizes Poulantzas, saying: "He erred in assuming

that somewhere in the state there is something which can somehow guarantee bourgeois class domination."" But while Poulantzas's language certainly does suggest some notion of such a guarantee ("the general negative limits of this autonomy"), it is also possible to interpret this claim not as an empirical prediction, but as a statement of tendency-that "in the long run" the workings of the state tend to reproduce class domination. On the interpretation of relative autonomy J am suggesting, the state is relatively autonomous in that it is both irreducible to economic relations and importantly determining and determined in relation to them. The economy and the state differ not simply in their personnel, but in their principles of existence and constitutive rationalities. The economy is constituted by a multitude of private institutions (enterprises, corporations, banks) whose rationality is particularism and private profit. The state is constituted by a set of public institutions (executive, parliaments, courts, police, armies,

administrative bureaucracies) whose rationality is the provision of internal and external order and "security" and the "general interest." The concept of relative autonomy denotes the structural relation between class relations and the state. Each is a condition of the others' existence, and each is in this sense implicated in the other, while at the same time maintaining its own respective autonomy. Because of this essential and necessary relation between state and class, because the state is what it is in virtue of its rela"Ibid., pp. 133-34. "Ibid., p. 136.

166

Marxism and Power

tionship to class relations, and vice versa, they are only relatively autonomous. The concept of relative autonomy thus designates the causal determinations between two social structures, state and relations of production, each of which is characterized by institutional specificity and causal effectivity. This understanding of the separation under capitalism of the sphere of public power and the sphere of private property does not mean that these spheres are hermetically sealed from each other, but rather than they are distinct and mutually determining. It also does not mean that the economy is strictly "private," in the sense of free of state interventions within it. As Poulantzas insists, the concept of relative autonomy does not denote the liberal, "noninterventionist" state of competitive capitalism, but the more general structural separation of political and economic relations. The continual interfaces of state and economy, and the extensive intervention of the contemporary welfare state in the economy, are in this view "transformed forms" of the relative autonomy of the capitalist state, which is a causal determinant even when it does not intervene, and which maintains its specific auronomy even when it does. 46 This interpretation of relative autonomy should be distinguished from a number of other recent interpretations. According to Fred Block:

The basic problem of "relative autonomy" is the conceptualization of the ruling class. Relative autonomy theories assume that the ruling class will respond effectively to the state's abuse of that autonomy. But for the ruling class to be capable of taking such corrective actions l it must have some degree of political cohesion, an understanding of its general interests, and a high degree of political sophistication .... Yet if the ruling class or a segment of it is class conscious, then the degree of autonomy of the state is clearly quite limited. At this point the theory of relative autonomy collapses back into a slightly more sophisticated version of instrumentalism. State policies continue to be seen as a reflection of inputs by a class conscious ruling class. 47 46Poulantzas, State, pp. 166-67. 47Pred Block, "The Ruling Class Does Not Rule," Socialist Revolution 7 (May-

June 1977), 6-28.

Marxism and the State

167

For Block, relative autonomy thus refers not to a structural relation, but to the relation between state institutions and a concrete group, "the ruling class." In this view the state is relatively autonomous (rather than fully autonomous) because it is ultimately manipulated as an instrument of the capitalist class organized as a ruling class. Block in fact opposes this collapsing of relative autonomy into class instrumentalism. In a more recent essay he suggests that the problem of relative autonomy points to the following questions: "What is the degree to which politics and the state have independent determining effects upon historical outcomes? Can the state or the people who direct the state apparatus be historical subjects?"4' He answers the second question in the affirmative, and concludes that state managers are hisrorical subjects. What is crucial here is that he interprets the question of relative autonomy as a question of subjectivity and agency, insisting that the objectives of some group, and not a structural relation, lie at its heart. An observation by Laclau is pertinent here: From the Poulantzas viewpoint this relative autonomy would be in turn a structural eiement, that is to say, the result of a particular articulation between the instances corresponding to the mode of production under consideration; in that sense, one more objective determination of the system as a whole. The relative autonomy of the state and its objective determination would be incompatible only if this autonomy were understood as a break in the chain of necessity and the emergence-however relative---of a realm of freedom. But this contraposition only makes sense within a problematic of the subject, which Poulantzas excludes by definition. 49

As I have remarked, Poulantzas's polarization of structure and agency is unwarranted. But this is equally true in the case of Block. To suggest that the concept of relative autonomy designates the articulation, or causal relationship, between class domination and the state, is not in the least incompatible with the agency of class actors arid state officials, both of which exercise their structurally 48Pred Block, "Beyond Relative Autonomy: State Managers as Historical Sub~ jects," in Ralph Miliband and John Saville, cds., The Socialist Register 1980 (London: Merlin Press, 1980), p. 227. 49Laclau, 1977, p. 65.

168

Marxism and Power

determined powers conjuncturally, within the framework of their particular definitions of reality. The concept of relative autonomy, however, marks not a moment where structural determination ends

and agency begins; it marks the structural delimitation of state power and class power. Miliband, in Marxism and Politics, provides a more subtle interpretation of the concept. According to him, relative autonomy refers to the "structural constraints which do beset any government working within the context of a particular mode of production."so On this view the concept does not mark the free will of a historical subject; it marks a relationship of structural determination, more specifically, the structural economic limits upon state action. This interpretation is closer to the one we are proposing. It suffers, however, from a one-sided emphasis on the structural determination of the state rather than on the determination of economic relations by the state. Miliband notes this in a more recent essay: "There has been considerable discussion among Marxists and others about the nature of the constraints and pressures which cause the state to scr,e the needs 01 capital-notably whether these constraints and pressures were 'structural' and impersonal, or produced by a ruling class armed with an arsenal of formidable weapons and resources. But beyond the differences that were expressed in these discussions, there was also a fundamental measure of agreement that the state was decisively constrained by forces external to it. ,,51 Miliband correctly insists upon the importance of the causal effectivity of the state. But astonishingly, he completely ignores a body of work within the Marxian tradition, particularly that of Poulantzas, which addresses precisely this point. And in doing so he endorses Theda Skocpol's claim that Marxists have not adequately considered "what states are in their own right. "52 As we shan see in the next section, the analysis of the state's role, as an autonomous institutional structure, in constituting and reSOMiliband, Marxism, pp. 73-74. He writes: "A capitalist economy has its own 'rationality' to which any government and state must sooner or later submit, and usually sooner." 51Ralph Miliband, "State Power and Class Interests," New Left Review, no.

138 (March-April 1983), 58-59. 52Theda Skocpol, States and Social Revolutions (Cambridge: Cambridge University Press, 1979), p. 29.

Marxism and the State

169

producing class domination, has been an important theme in recent Marxian writing. The concept of relative autonomy, in this regard, is best construed as denoting the mutual determinations between class relations and the state.

The State as a Determinant of Class Relations Marxian theorists have focused on a numher of ways in which the capitalist state secures the conditions of capital accumulation and maintains the domination of capital. These state functions, or effects of state power, have as their not necessarily intended consequence the tendential reproduction of class relations. We emphasize the unintentional nature of these reproductive functions insofar as most of the theorists we will discuss posit no teleology, or purposive mechanism, which pursues these consequences. The effects are tendential in that the theory of the state posits no guarantee that the state will actually reproduce class relations; it only posits that the state tends to have this effect because of its specific mode of operation. Thus, despite the interest of certain recent Marxists in functionalist explanation, the explanatory schema I will outline below is not to be understood in functionalist terms. 53 Insofar as we employ the language of "functions," there is no implication of teleology or iron-clad necessity; the functions of the state are simply those things the state does, the typical effects of state power. The first function emphasized by Marxian theorists, and the most obvious, is the state's narrowly legal function. The capitalist state provides a uniform set of standards, a rationalized code of civil law, which makes capitalist activity possible, both by defining private property rights (contracts, torts, liability, inheritance, etc.) and 53